Mediated Bricolage and the Sociolinguistic Co-Construction of No Sabo Kids

Abstract

Sociolinguistic styles and the resultant ascribed identities are understood as the product of simultaneous variables, leading to the notion of bricolage, or the co-occurrence of variables and their collective indexical meanings. Relatively little attention has been paid to these processes as they manifest on social media platforms. The goal of the current paper is to understand which linguistic and thematic features co-occur in the online production of the no sabo kid style and identity, which manifests as a form of linguistic discrimination towards U.S. Latinx youth. “Hashtag communities” were used to locate posts about no sabo kids on TikTok (N = 95), and videos were automatically and manually coded for salient linguistic and discursive resources in the online no sabo kid community. The results show the co-occurrence of code-switching and phonological and lexical variation, alongside discursive themes, namely ‘proficiency’, ‘ethnicity’, and ‘performative lexical gaps’. I argue that the no sabo kid hashtag community is a mediated manifestation of ideologies surrounding U.S. Latinx bilinguals, where a supposed lack of proficiency in Spanish and grammatical blending of Spanish and English index inauthentic ethnicity. Mediated instantiations of sociolinguistic styles shed light on how linguistic features become enregistered through multimodality and semiotic bricolage.

1. Introduction

The same linguistic output can often have different social interpretations. In the context of child language acquisition, sabo (I know) in Spanish is an overregularization of an irregular present tense verbal inflection (saber.INF > sé.PRES.1PS). Overregularizations like these are common among children acquiring Spanish (Clahsen et al. 2002), and, when produced by young children, are often met with laughter and reflect affective indexes of ‘cute’, ‘playful’, and ‘childlike’. Yet, when the same overregularization is produced by an adolescent or young adult heritage speaker of Spanish, the evaluative reactions can shift to those of ‘ridicule’, ‘embarrassment’, ‘lacking Spanish proficiency’, and ‘inauthentic ethnic identity’. The purpose of the current paper is to critically address the latter examples, whereby U.S. Latinx social media users produce linguistic features associated with a no sabo kid persona. Stransky et al. (2022) note that the identity label is used in mediated spaces to make reference to Latinx youth who are dominant in English and show evidence of language mixing (i.e., borrowing, calquing), all the while functioning within a “linguistic oppressive movement that openly criticizes US Spanish and its speakers” (p. 38).

To date, extensive research has been conducted on the ideological dimensions of heritage Spanish communities and speakers (e.g., Leeman 2018; Rosa 2019), and an area of growing interest is how linguistic ideologies manifest in ethnolinguistic identity production, particularly within new media. As such, the present research directly addresses a call for greater disciplinary connection between linguistics and media studies (Sánchez-Muñoz and Retis 2022) and aims to provide a detailed expansion of Stransky et al.’s (2022), which brings awareness to the current issue and highlights the marginalizing forces of the term no sabo kids in social media videos.

1.1. U.S. Latinx Language and Ideology

Despite ongoing political efforts to make English the federal official language of the United States (U.S.), other languages are well represented: nearly one quarter of all school children in the U.S. speak a non-English language at home, and the number of students classified as English Language Learners is likely to increase by 50% by the year 2025 (Potowski 2021). Historically, the Spanish language has been spoken on land that today we consider the U.S. for longer than English, and currently U.S. Latinxs1 constitute the country’s second largest racio-ethnic population, with U.S.-born Latinxs outnumbering those born abroad (Funk and Lopez 2022). Between 1980 and 2019, self-rated English proficiency among U.S. born Latinxs increased by 19% and, conversely, use of Spanish in the home decreased by 10% (Funk and Lopez 2022).

This demographic information gives rise to questions about prestige attributed to the Spanish language in the U.S. context. In South Florida, where Spanish is narrativized as a language of necessity, studies show overwhelming prestige and biases in favor of English as compared to Spanish (Callesano and Carter 2022; Carter and Lynch 2014). In the U.S. Southwest, Rangel et al. (2015) show that, in terms of “status”, “solidarity”, and “personal appeal”, Spanish and English are rated equally, while code-switching is rated lower than both monolithic understandings of the two languages. These examples reflect colonial and monolingual language ideologies (e.g., Wiley 2014), which set the stage for a psychosocial outcome—linguistic insecurity among minoritized language communities.

Linguistic insecurity among heritage language speakers is well documented, particularly in the context of heritage language education (e.g., Martinez and Petrucci 2004). However, outside of the context of education, the negative feelings that heritage speakers often experience towards their own linguistic abilities in the minoritized language—Spanish in this case—is of key importance for sociolinguistic questions of indexicality and style, as these attitudes mediate linguistic performance and use. In a study of linguistic insecurity, Tseng (2021) argues there is an “imposed deficit” attributed to second and third generation Latinx bilinguals, which leads to negative effects on their “linguistic self-identities and self-esteem” (p. 131). The everyday lived experiences of U.S. Latinx bilinguals can be described as facing an ideological double bind—racialized as low proficiency English speakers (e.g., Flores and Rosa 2015) and judged as not proficient enough in Spanish. The process of linking linguistic practices with racial/ethnic categories emphasizes the creation of difference and deficit, allowing for the subordination of multilingual, Latinx linguistic repertoires against monolingual and white language practices (García et al. 2021). The linguistic policing that happens has been described as “linguistic terrorism” in Anzaldúa (1987) and Christoffersen (2019). Bilingual speakers face hostile perceptions and actions, like unnecessary repetition requests (as the data here will show) and placement in remedial education programs, for example. The racialization processes that lead to linguistic insecurity also reflect how shame is a common outcome of linguistic socialization (Lo and Fung 2011).

Whether the perceptions that are studied among U.S. Latinx communities attend to “Spanish” and “English” as conceptual categories (e.g., Kutlu and Kircher 2021) or to specific linguistic features (e.g., Chappell 2019), language-based social biases persist within socio-historical systems that allow them to manifest in material ways. For example, Carter and Callesano (2018) note that different national dialects of Spanish (Cuban, Colombian, and Peninsular) are associated with different levels of occupational prestige and socioeconomic status. African American English (AAE) has long been studied in terms of materialized perceptions, for example in the workplace (McCluney et al. 2021), the court system (Rickford and King 2016), and automatic speech recognition programs (Martin and Wright 2023). As such, the perception of minoritized linguistic varieties—and the bodies that produce them—materialize as not eligible for higher paying employment, not credible in legal environments, and disadvantaged within technological systems, underscoring the ways in which marginalized Latinx bilingual language practices are policed.

This policing takes place within the contexts of standard language (Lippi-Green 1997), monolingual (e.g., Wiley 2014), and raciolinguistic ideologies (e.g., Rosa and Flores 2017)—the axes to which speakers and listening subjects (Inoue 2003) orient themselves when enacting, performing, or evaluating ethnolinguistic identities. To that point, we now turn to a discussion of indexicality, enregisterment, and style.

1.2. The Co-Construction of Ethnolinguistic Identity: From Indexicality to Performance

There is no linguistic variable that occurs outside of a social environment and the social meanings that are attributed to linguistic resources are fluid according to context and listener. Sociolinguists have collectively used the notions of indexicality (e.g., Silverstein 2003), enregisterment (see Agha 2003; Johnstone 2016) and bricolage (Eckert 2008; Guy and Hinskens 2016; Hebdige 1979, among others) to think through the ways in which semiotic resources (i.e., linguistic variables, music, clothing) come together in the production of social meaning. On the one hand, indexicality refers to how linguistic variants point to certain social traits, characteristics, and qualities, such as how variable realizations of syllable-final /s/ in Spanish may index—or perceptually indicate—various levels of social status (e.g., Chappell 2019). Enregisterment, on the other hand, refers to another stage in the processes of assigning social meaning to language features, where linguistic forms ultimately point to identifiable social beings. Johnstone (2011) describes how a set of phonetic and morphosyntactic variables collectively signal a Pittsburgh identity, such as “yinz” as a second-person plural pronoun, multiple negation, and monophthongization of /au/. In Pittsburghese and other language varieties, different linguistic variables are known to co-occur with one another, forming the building blocks that collaboratively index or enregister some social outcome, such as a quality, attribute, or a named social being. The co-occurrence of multiple linguistic variables and their indexical meanings is known as bricolage (Eckert 2008; Guy and Hinskens 2016; Hebdige 1979).

Furthermore, all these processes function within ideological systems about language (Schieffelin et al. 1998), meaning that those individuals and groups who exist at the privileged hegemonic center have the power to dictate, in part, the social meanings that are assigned to language. Individuals whose linguistic repertoires match those of the ostensibly socially privileged are not deemed deficient and, instead, gain capital that is tied to language (Bourdieu 1991). These processes are explored in what Eckert (2012) refers to as third-wave sociolinguistics, which seeks to elucidate how language users take advantage of the variable social meanings that are attributed to linguistic features as they perform various versions of themselves.

To exemplify the theoretical points above, I turn to Mendoza-Denton’s (2008) seminal research on language and identity in two Latina youth gangs in a California high school. On the one hand, the Norteñas are “Americanized”, and to show that, they forefront their bilingualism through translingual practices. On the other hand, the Sureñas orient themselves more towards Latin America and thus use Spanish more frequently. The “Norteñas ‘feel’ themselves Chicanas, Sureñas ‘feel’ themselves Mexican, and are both opposed to and in intense competition with each other in every aspect of their symbolic existence” (Mendoza-Denton 2008, pp. 250–51). One way this competition becomes clear is through semiotic resources. The Norteñas prioritize red and burgundy for their clothing and lipstick, have feathered hair, and listen to Oldies, which goes hand-in-hand with their prioritization of the English language. The Sureñas prefer blue and navy clothing with brown lipstick, wear ponytails, and listen to banda music, which is then linked with their more frequent use of Spanish. In this one example, we can see how indexicality and bricolage come together, where the social meaning of a particular linguistic repertoire is reinforced by various semiotic resources building upon each other. Furthermore, paying attention to a specific phonetic variable, Mendoza-Denton’s (2008) results show that the Norteñas and the Sureñas do not differ significantly in their production of raised /ɪ/2 and as such, the indexical values associated with raised /ɪ/ are dependent upon the simultaneous realization of extralinguistic features—or, in other words, the co-occurrence of sociolinguistic variables.

Numerous studies have been conducted on the co-occurrence of linguistic features within ethnolinguistic varieties (e.g., Newlin-Łukowicz 2016; Oushiro 2016), but studies of a similar nature are still lacking when it comes to Spanish, particularly for U.S. Latinx bilingual speech communities. Among the few that exist, Erker and Otheguy (2016) examine the co-occurrence of pronoun use, coda /s/, subjunctive morphology, pronoun placement, and word order in New York City and argue for effects of social saliency and Spanish-language ideologies. The aforementioned studies tend to take a variationist sociolinguistic approach, quantitatively exploring how and when variables co-occur, while the data shown below, however, are understood through the notions of linguistic style and mediated performance. When working with data from performance, which are not naturalistic in the early variationist sense, it is useful to note that some videos have a clearly parodic goal and others are simply recordings of interaction. Whether in a performative context or not, linguistic variables are not only produced alongside other linguistic resources (e.g., discourse topic) but also other semiotic materials (e.g., particular clothing, food, or music choices).

In the subsequent analysis, I specifically combine Rampton’s (1999) stylization, Johnstone’s (2011) linguistic performance, and Zappavigna’s (2015) hashtag community in order to place the social media videos in a performative and community-centered context. Rampton (1999) argues that semiotic variables are used as tools in the production of a linguistic style and that these tools then become enregistered with a person-type. Here we can think of the linguistic styles of U.S. presidents (Slatcher et al. 2007), broader styles of a European multiethnolect (Quist 2008), or “Valley Girls” versus “Ghetto Girls” (Goodwin and Alim 2010), for example. According to Johnstone (2011), linguistic performances of enregisterment hinge on paying “meta-communicative attention to [themselves], putting on display not only what the message means but how” (p. 676). Research on linguistic performance illustrates how various semiotic variables co-occur within mediated spaces (e.g., Calhoun 2019; Chun 2004; Slobe 2018). For example, parody performances on YouTube show how collections of semiotic features like uptalk, tag questions, and blond hair work together to enregister the Mock White Girl persona, but also how the performers’ ideological stances towards the persona vary (Slobe 2018). Additionally, Chun (2004) demonstrates how the Mock Asian style in standup comedy performances, which features, for example, confusion of liquids and prayer bows, depends on and reproduces “racist ideologies of Asian othering and ridiculing” (p. 287). Both stylization and linguistic performance can be conceptualized in the context of mediatization, and one mediated place where we can locate examples of stylized performance abound is the hashtag community—a mediated community whose membership is determined by using and interacting with metadata, such as a hashtag (Zappavigna 2015). Thus, #nosabokids functions as a mechanism for online community organization. Social media users can choose to incorporate hashtags into their posts for several reasons, such as making the post searchable and/or providing labels or metacommentary. The myriad combinations of features, which can be found within a hashtag community, lead to embodied styles, and the data presented here aim to illustrate the sociolinguistic processes through which the no sabo kid identity is performed and sustained on social media.

2. Materials and Methods

The materials and tools used for the current analysis are Python, RStudio (R Core Team 2022), and publicly available videos on the social media app, TikTok.3 The app has gained immense popularity in recent years and data from 2021 show that 21% of people in the U.S. use the app. Nearly half (48%) of all U.S. adults between the ages of 18 and 29 say they use the app (Auxier and Anderson 2021), and the data discussed here are understood within the context of mediated spaces associated with younger adults at the crossroads of the Millennial generation and Generation Z (Dimock 2019).

The data collection and coding process consisted of two stages. First, a Python script designed to access TikTok’s API (Teather 2022) was modified to fit the parameters of the study (see Brown 2023). A total of 118 videos that included one of the four hashtags—#nosabo (n = 30), #nosabokids (n = 30), #yonosabokid (n = 28), #nosabokidsbelike (n = 30)—were automatically collected and coded (see

Table 1. Codes for information gathered from Python script.

Python Code	Output
ID	numerical 1–118
URL	link to TikTok video
video_author	username of video producer
hashtags	list of hashtags
caption	caption text
isAd	true/false if advertisement
stickersOnItem	text of sticker(s)
privateItem	true/false if private video
commentCount	count value of comments
playCount	count value of plays
shareCount	count value of shares
searchQuery	hashtag used in search
isPairedWith	true/false if paired with #mexico

for codes). As of 12 December 2022, the hashtags listed above had the following numbers of views on TikTok: 440.9 million, 242.4 million, 937,000, and 47.2 million, respectively, and all video engagement data reported in this article are based on that date. The total runtime for the Python script was 8 min and 30 s on a 2019 Macbook Pro. After collecting this information, a post-hoc analysis was run to see whether the hashtag of interest was found alongside #mexico within the same video caption. This was motivated by the observation that numerous lexical indexes used in the videos are stereotypes of Mexican Spanish (as shown below). By querying whether #mexico is used alongside the other hashtags, results might suggest that the videos producers are taking a stance to ideologically align the no sabo kid identity with a particularly Mexican heritage.

During the analysis stage, 23 videos were removed from the dataset due to irrelevance to the subject, repetition because of co-occurrence of two or more hashtags, the video’s topic being on Anglo-white second language learners of Spanish, the video having been deleted from the user’s profile, or the account having changed its settings from public to private. After all exclusions, the final analysis consisted of a total 95 videos produced by 65 users.

The second stage of coding was conducted manually and included the information outlined in

Table 2. Manual coding criteria for each video.

Manual Code	Output
phonetic	presence (1), absence (0)
lexical	presence (1), absence (0)
code-switching	presence (1), absence (0)
theme	grounded qualitative coding
type	parody (p), naturalistic (n), commentary (c)

. This stage is divided into three structural categories—phonetic, lexical, and code-switching. Phonetic codes refer to instances of bilingual sound production, lexical codes pertain to words or lexicalized phrases, and code-switching notes any discourse where Spanish and English are simultaneously present. First, each video was coded for the presence or absence of code-switching, as well as phonetic and lexical variables. After the initial round of coding, the lexical category was divided into two sub-categories: nominal (e.g., semantic extensions) and verbal (e.g., morphological blending). These categories were used based on impressionistic data from a pilot analysis conducted in 2021, where videos were qualitatively analyzed for the saliency of linguistic variables employed in the production of the no sabo kid identity. Additionally, the codes in Table 2 are founded in previous research on sociolinguistic bricolage and enregisterment (e.g., Johnstone 2011; Newlin-Łukowicz 2016).

For example, let us say we found the following features in one video: production of /r/ as [ɹ] in the word gracias [gɹa.sjas], examples of interactional code-switching between the speaker and an interlocutor, and also an example of a lexical play on words, such as pedo ‘fart’ for puedo ‘I can’. This example would be coded as 1 for phonetic, 1 for lexical, and 1 for code-switching.

Next, the theme of each video was determined using inductive qualitative coding, specifically following Bingham and Witowsky (2022), where emerging themes were coded by taking into account the entirety of the video—discourses, stickers, sounds, captions, and hashtags. The themes of ‘performative lexical gaps’, ‘proficiency’, and ‘ethnicity’, arose from within the data, but also follow previously established sociolinguistic theory regarding this particular ethnolinguistic community (e.g., Tseng 2021). One video can have multiple themes; however, for the purposes of this analysis, one theme was selected as the most prominent. Lastly, each video was coded for type; distinctions were made if the video was clearly a parody (n = 39), naturalistic (n = 15), in the sense that the recording was made of a genuine interaction with no intention to act or perform, or commentary (n = 41), where the video either consisted of a monologue style of meta-linguistic commentary about no sabo kids or the video was created using a meme style, where text (the stickersOnItem in Table 1) was overlaid on top of a short clip with the purpose of that text being the commentary.

When working with data from social media, and considering speakers4 within the hashtag community, taking note of views, likes, shares, and comments helps to understand community scale, that is, how widespread the hashtag community is and how much other people engage with the material. See

Table 3. Corpus engagement statistics (all videos).

Count	Range	Mean	Standard Deviation
Play (number of views)	Min. 513 Max. 12,600,000	1,402,634	2,494,473
Share (number of shares)	Min. 0 Max. 125,100	9145	21,521
Comment (number of comments)	Min. 0 Max. 25,400	1834	3619

for this information across the dataset. The corpus of videos shows a wide range of plays, shares, and comments, indicating that videos within the no sabo kid hashtag community circulate frequently throughout TikTok. It is an online community that reaches millions of views, which shows that the no sabo kid persona is prolific on social media.

3. Results

The data presented here speak to the specific linguistic and discursive variables that are employed within the no sabo kid hashtag community on TikTok. In the sections that follow, I describe the indexes at a phonological level as well as the indexes considered to be lexical, which are divided into nominal and morphological categories. Then, I consider their co-occurrence alongside specific discourse themes. I note that nearly 60% of all videos demonstrate code-switching. I also argue that code-switching functions as yet another gear in the mechanism of co-constructing the no sabo kid identity, following studies showing the lack of social prestige of language mixing in light of monolingual norms (e.g., Rangel et al. 2015) and how environments including both Spanish and English are overtly labeled as “ghetto” as a result of the racialization of language in southern California (Bucholtz et al. 2007). The examples provided below demonstrate the linguistic layering in the construction of a derogatory, ethnolinguistic persona.

3.1. Phonetic Indexes

Of the 95 videos, 31 included at least one instance of a phonetic production that indexes a no sabo kid identity. While all examples are indicative of either phonetic or phonological differences between Spanish and English, numerous cases involved different instances of rhotic variation. Other examples demonstrate consonant cluster reduction, affrication, and differences between the English and Spanish vocalic systems. Of note is that all the Spanish phonetic realizations that index no sabo kids do so vis-à-vis English. Whether intentional through parody or as a part of spontaneous speech, by inserting English-like phonetic realizations into Spanish discourse, the index signals proficiency in English, which has as its byproduct a perceived lack of proficiency in Spanish.

Furthermore, and to the argument for the co-occurrence of these variables within and across the social media posts, it is common to find these phonetic features couched within lexical indexes of the no sabo kid identity.

Table 4. IPA of phonetic examples.

Non-Stigmatized Production	Production Indexing No Sabo Kids	Example Context
[ɾ]	[ɹ]	señora [se.ɲo.ɹa]
[r]	[ɾ]	perro [pe.ɾo]
[ɾ]	[r]	fuera [fwe.ra]
[ɾd]	[r, l]	perdón [pe.ron] or [pe.lon]
[t]	[tʰ]	taco [tʰa.koʊ]
[dɾ]	[$\stackrel{⏜}{\mathrm{d}\mathsf{ʒ}}$]	madre [ma.$\stackrel{⏜}{\mathrm{d}\mathsf{ʒ}}$ɹe]
[we]	[e]	puedo [pe.ðo]
[o]	[oʊ]	sabo [sa.boʊ]

shows the phonetic indexes that are most present in the dataset, highlighting bilingual productions where monolingual understandings of Spanish and English phonetics are set apart from one other. Transcript 1 illustrates an example of a phonetic index being mobilized to label someone as a no sabo kid.

Transcript 1. #nosabokidsbelike video with 150 comments, 137,300 plays, and 49 shares

1	A: A ver, Jimena, ¿cómo se preparan los tacos?
2	‘Let’s see, Jimena, how are tacos prepared?’
3	A: Enséñanos.
4	‘Show us’.
5	A: ¿Cómo se preparan los tacos?
6	‘How are tacos prepared?’
7	B: ‘Esto es carnita’.
8	This is meat.
9	A: You put the carnitas [kaɾ.ni.tas] in the tortilla [toɾ.ti.ʝa]
10	‘You put the meat in the tortilla’
11	B: uh huh
12	A: mmk. And then?
13	B: Hold on.
14	B: *walks over to another table with toppings*
15	B: cilantro! [sə.lan.$\stackrel{⏜}{\mathrm{t}\mathsf{ʃ}}$ɹo]
16	A: huh!?
17	B: cilantro! [sɪ.lan.$\stackrel{⏜}{\mathrm{t}\mathsf{ʃ}}$ɹo]
18	A: cilantro [sɪ.lan.$\stackrel{⏜}{\mathrm{t}\mathsf{ʃ}}$ɹo]
19	A: No ss.
20	You a no sabo kid?

To contextualize, speaker A is recording the video and the viewer is limited to only hearing their voice. Speaker B is a young adult, wearing a University of California, Los Angeles sweatshirt. She is inside of a home where a party or gathering is taking place and person A is recording her as she prepares her plate of food. Based on the hashtags used in the video caption, we assume this interaction took place in Staten Island, New York (#statenisland). Speaker A uses phonological code-switching in line 9, producing the food items with Spanish phonology. Then, in line 15, speaker B realizes ‘cilantro’ with an English-like phonology (i.e., reduced and centralized vowel, affrication of <tr>). This is met with a jarring “huh!?” by speaker A (line 16), which prompts a repetition by speaker B. Speaker A then reiterates the word, as if to mock speaker B in line 18. Both speakers appear to be proficient bilinguals, yet it is speaker B who is being video recorded, indirectly asked to validate her Latinx identity via the process of putting together a taco, and whose linguistic production of ‘cilantro’ is mocked and used as the trigger for the no sabo kid identity. Crucially, as will be discussed further, the co-construction of the no sabo kid identity is not just the work of the speaker, but that of the speaker and the listener (D’Onofrio 2015; Rosa and Flores 2017), which here includes a multitude of ambient listeners on social media. Additionally, the phonological alternation across speakers A and B reflects an issue of bilingual familiarity, discussed by López (2020) and Delgado (2020), where words with “common currency” between the two languages can be realized using either phonological system, but the pronunciation of words like carnitas, tortilla, and cilantro with particular phonological patterns does socio-political identity work (Hill 2009). It seems to be the case, then, that there is a co-occurrence of at least two types of linguistic indexes in transcript 1, phonetic and lexical.

3.2. Lexical Indexes

The lexical indexes are divided into two categories: nominal and morphological. Approximately 25% of the videos include a nominal example of a lexical index, and lexical entries coded as nominal include three sub-types: (1) plays on words via phonetic similarity, (2) lexical stereotypes of Mexican Spanish, and (3) phonological adaptation and calquing.

On the one hand, nominal plays on words occur within Spanish and two examples are: noches/nalgas ‘nights’/‘butt cheeks’ and puedo/pedo ‘I can’/‘fart’. On the other hand, some are cross-linguistic, such as false cognates (sopa ‘soup’ referencing ‘soap’ and embarazado ‘pregnant’ referencing ‘embarrassed’). There are other examples of more creative false cognates, such as veterinario ‘veterinarian’ used for ‘vegetarian’, and also inverted Spanglish (Rosa 2016)—addresso ‘address’—showing the addition of morphological -o to an English nominal root. These plays on words categorically occurred in the parodic type of TikTok video, indicating the hyper-performative nature of these indexes. To that point, it is likely that the comedic effect of phrases such as buenas nalgas instead of buenas noches or no pedo for no puedo comes from the rupture of the pragmatic frame, where listeners (viewers on TikTok) have an expectation of a linguistic exchange, and when that expectation is not met, the frame is said to be broken and a possible result is humor (Goffman 1974).

Across the set of videos analyzed here, examples of Mexican Spanish lexical stereotypes are: guëy ‘man’/’dude’, no mames ‘no way’, a la verga ‘fuck it’, que hubo ‘what’s up’, and compa ‘friend/mate’. These lexical items are thought of as stereotypes in the Labovian sense, where they are not only used in stylistic variation but are also discussed in meta-linguistic terms (Labov 1971). 55% of the videos in the dataset included #mexico, suggesting an explicit connection between the no sabo kid persona and Mexican heritage.

Finally, the category of phonological adaptation and calques consists of examples where either a lexical form in English is phonologically adapted to Spanish or a structure from English (e.g., skyscraper) is produced as a literal translation in Spanish (e.g., rascacielos). Examples of the former are the productions of gualmar [wal.maɾ] for English ‘Walmart’, and an example of the later is chi-chi de pollo for ‘chicken breast’, which uses a colloquial term for women’s breasts in place of pechuga, the Spanish word for ‘breast’ in the context of food. The data also show evidence of lonjas [lon.xas] for English ‘lungs’, a semantic calque with some cross-linguistic phonological similarity. It is important to keep in mind the performative nature of these examples; phrases like chi-chi de pollo are unlikely to become part of the linguistic norm of the speech community, contrasting with phonologically adapted forms like gualmar.

The data show that the lexical indexes, whether nominal or morphological, are (a) not very frequent and (b) are found primarily in the parody and commentary video types, with just one example falling in the naturalistic category (see Figure 1). From the morphological category, examples of morphological blends overwhelmingly show English verbal roots with Spanish inflection, which aligns with Toribio and Bullock’s (2016) findings from Spanish heritage speakers in Texas. The examples in

Table 5. Examples of morphological blends.

Verbal Category	Morphological Blend	Interpretation
Progressive	barkiando	barking
	jumpiando	jumping
Infinitive	danciar	to dance
	parkiar	to park
	catchar	to catch
	askiar	to ask
	helpiar	to help
	workiar	to work
	blamiar	to blame
	popiar	to pop
	cutiar	to cut
Finite	blamias	blame [+present, 2nd person singular]
	tripié	trip [+past, 1st person singular]

present the orthography that was used by the TikTok producers.

In the current dataset, there is exactly one instance of a morphologically blended index of the no sabo kid identity within the naturalistic video type category (see Figure 1). In Transcript 2, a child of approximately 4–6 years old is being recorded by an adult (assumed to be her mother or caregiver). While the young child is aware of the recording that is happening, the child is giving the adult directions about the recording placement and method, indicating a non-performative moment. The video producer (assumed to be the adult the viewers do not see and the one who posted the video to TikTok) turns the purpose of the video into a meta-linguistic commentary about the child’s use of ‘grabbing’. The bilingual child likely retrieved the English verbal root “grab” from the Spanish infinitive grabar ‘to record’ and attached to it the English progressive marker -ING. The caption of the video is, “She tries! She’s so cute!” with two heart-type emojis and a laughing-with-tears emoji, and includes the following hashtags: #mexican, #latina, #cute, #nosabo, #nosabokidsbelike, and #texas. The video producer, speaker A (adult), placed a sticker (S) on top of the video (Line 1), which is spoken by an app-internal voice and heard by the viewer. Speaker C represents the child.

Transcript 2. #nosabokidsbelike video with 374 comments, 472,900 plays, and 974 shares

1	S: No sabo kids be like
2 3	C: Stay grabbing here *eye gaze towards the floor and open hand gesture to suggest to the interlocutor they should not move* ==
4	A:    == What’s grabbing?
5	C: *Steps closer to camera and phonates something like [ps:]
6	C: Just keep (..)
7	Wait for me! ==
8	== *raises hand in open palm gesture to interlocutor*
9	But stay the camera there.
10	A: Ok

3.3. Thematic Indexes

As a part of the manual coding process, each video was coded for theme via an inductive approach. While many themes became apparent throughout these videos, I focus here on the three most common. The third most frequent theme was that of ‘proficiency’, (13%) with ‘ethnicity’ as the second most common (16%), and finally, ‘performative lexical gaps’ as the most common (20%).

In one proficiency-related video (2 comments, 513 views, and 0 shares), which includes the caption “insecure speaking spanish as if it wasnt my first language” and the hashtags #latina, #yonosabokid, #spanish, and #bilingual, a young adult relates that she watched a comedy skit online, and many written comments claimed the comedian had a “yo no sabo accent”. The speaker of the TikTok video analyzed here makes a rhetorical move to say that she has no hate towards no sabo kids and that one of her friends is one, but if someone were to refer to her as a no sabo kid, she would cry. She moves to say that she is already the “whitewashed” one among her cousins and that is because she is “not like Cuban, y’know”, with phrasal stress placed on “Cuban” to suggest that she is Cuban, but not quite as Cuban as her cousins. Next, she says it would be her figurative “last straw” if someone called her a no sabo kid and then the video cuts to the same speaker reading very brief segments of a book aloud, where she performs a lack of reading ability in Spanish.

In another video, one that takes the form of a commentary type, the factor of ethnicity is much more direct. In just a seven second video, with 36 comments, 13,200 plays, and 73 shares, the overall argument is that if you do not speak Spanish, you cannot be Dominican. The video sticker reads, “Cuando me dicen que son dominicanos pero no saben hablar español” ‘When they tell me they are Dominican but they don’t know how to speak Spanish’, and in the video a person is sitting down within a facial gesture indicating confusion. The caption reads, “SI USTED ES DOMINICANO Y NO HABLA ESPAÑOL UTE NO E DOMINICANO NA !!” ‘IF YOU ARE DOMINICAN AND DON’T SPEAK SPANISH, YOU ARE NOT DOMINICAN AT ALL !!’. The message here is that to be Dominican, speaking Spanish is required, according to this TikToker, and the link between language and ethnicity is reinforced in this video in two ways: first, in the interaction between the video’s sticker and the facial gesture of the user and second, in the caption. Of note is how the producer of the video actively switches into a more Caribbean, or Dominican, way of speaking through the written form of the caption. With “UTE NO E DOMINICANO NA” the producer is arguing that to be Dominican you need not just speak Spanish, but your Spanish needs to sound a certain way (i.e., omit /s/ and /d/).

Finally, regarding the most common theme of ‘performative lexical gaps’, I include a transcript of a parody TikTok, which uses a greenscreen of a nightclub and a sticker that reads “No Sabo Kids be like:” In this interaction, while there are three distinct characters, all are acted out by the same person. S is the main character in the skit (i.e., the no sabo kid), tía is the main character’s aunt, which makes a reference to women figures as sources of validation in Latinx families (Vasquez et al. 2022), and DJ is a background character in the context of the nightclub.

Transcript 3. #nosabo video with 2129 comments, 1,900,000 plays, and 6166 shares

1	Tía: Ven, mijo, ven ven ven!
2	‘Come here, mijo5, come come come!’
3	S: No no tía es que
4	‘No no aunt it’s that’
5	Yo no (.) ugh!
6	‘I don’t’ (.) ugh!
7	How do you say dance in Spanish?
8	No sabo (.) no sabo danciar
9	‘I don’t know (.) I don’t know how to dance’
10	Tía: *gasps* ~ sticker reads *shocked in Spanish* ~
11	¡¿qué dijistes?!
12	‘What did you say!?’
13	DJ: No mames, un pinche no sabo
14	‘No way, a fucking no sabo’
15	S: ¿Por qué todos me miran? Es que yo no sabo danciar
16	‘Why is everyone looking at me? I don’t know how to dance’.
17	No sabo danciar!
18	‘I don’t know how to dance!’
19	Tía: Mijito no se dice “danciar”
20	‘Mijito we don’t say “danciar” (to dance)’
21	Se dice bailar
22	‘We say “bailar” (to dance)’
23	S: Ah ok
24	p p p p perrón
25	‘p p p p perrón (a play between perrón ‘cool’ and perdón ‘sorry/excuse me’)
26	Tía: No te preocupes mijito yo te enseño cómo bailar
27	‘Don’t worry mijito, I’ll teach you to dance’
28	DJ: ¿Alguien dijo perrón?
29	‘Did someone say “perrón”?’
30 31	*music starts playing with the lyrics “algo que está perrón que toda la gente brinca” ‘something that is so cool that everyone jumps up’*

At the core of this parody is the supposed missing lexical entry for “to dance” in Spanish. To fill this ostensible gap, a bilingual can use their linguistic awareness to creatively come up with what might be a plausible outcome by adding Spanish verbal morphology to an English verbal root. Each and every instance of danciar is immediately preceded and followed by the emoji , indicating a mocking tone. In the clip we also see evidence of shaming from a familiar adult, a topic to be discussed further below, as well as other examples of sociolinguistic indexes of the no sabo kid identity, such as perrón for perdón. The video producer is aware of the play on words between perrón and perdón, as the song lyrics that are heard at the end of the clip use the term in a more appropriate context, as an adjective describing something as really ‘cool’. While this video is certainly parodic, evidence in lines 11 and 26 indicate experience with stigmatized sociolinguistic variants in Spanish—second person singular preterit /s/ (Barnes 2012) and heritage Spanish systems, where for example English “how” is mapped onto Spanish infinitives. This illustrates an interaction of stigmatized variables, some parodic and some not, in the mediated production of the no sabo kid.

4. Discussion

The evidence from this mediated approach to sociolinguistic style, enregisterment, and identity shows that phonetic and lexical variables, as well as code-switching and narrative themes collectively play key roles in the production (and performance) of the no sabo kid persona. While the language features noted here may index various attributes (e.g., friendly, trustworthy), the collection of variables enregisters the no sabo kid identity, or in other words, the social being to which the features point (Johnstone 2016). In her work on enregisterment and performance, Johnstone (2011) notes that audiences are asked to consider what it could mean that a specific character (e.g., a mother) has a specific accent (e.g., Pittsburgh) while also sounding working class (p. 676). Additionally, she notes that not all audience members/listeners will perceive the indexical relationships in the same way. In the data presented here, we consider what it could mean that a particular community (e.g., Latinx young adults) speaks a certain way (e.g., mixing Spanish and English phonetics, morphology, and discourse), while also facing pressures of racialization and linguistic subordination. Across naturalistic, commentary, and parody video types, the data show, for example, how rhotic variation is couched within lexical plays on words, how verbal roots in one language are combined with verbal suffixes in another, and how the bricolage of bilingual features are encompassed by themes of proficiency, ethnicity, and performative lexical gaps.

Whether through parody or not, enregisterment of the no sabo kid is accomplished through a clear mechanism—the hashtag community (Zappavigna 2014, 2015). Each video included in the analysis is overtly marked with the name of the social being to which the semiotic features point, namely #nosabokids (or any variation of the hashtag therein). While the phonetic, lexical, and discursive categories within this hashtag community were treated separately, all are in constant interaction. The hashtags themselves also function in conjunction with other hashtags, such as #mexico. Just over half of the videos in the dataset showed co-occurrence with #mexico, making a clear national origin identity link with the no sabo kid persona, though there is evidence of non-Mexican associations as well. Furthermore, the variables that index no sabo kids carry various sociopolitical weights, one of which is intergenerational shame.

Research on language socialization (Lo and Fung 2011) shows that shaming (often between parents and their children) is the result of local ideologies and sociocultural context and a “belief in the power of language” (p. 186). Furthermore, not all shaming is necessarily negative, and it can be paired with feelings of play. Yet, the interactional moves of parents, caregivers, teachers (i.e., adults in positions of power) to ask for repetition or imitate young Latinx bilinguals’ linguistic productions has the potential of internalizing shame that is directly linked to language. Following Rosa (2019) and Lawrence and Clemons (2022), when that shame is linked to language, it is simultaneously linked to race and ethnicity, raising issues of not just linguistic insecurity, but also authenticity.

Models of bilingualism, especially in the context of U.S. schools, are largely based on linguistic deficits (e.g., García 2002)—or what bilinguals supposedly lack. The video productions analyzed here, which come from an app used primarily by people between the ages of 18 and 29 in the U.S., can be thought of as manifestations of imposed sociolinguistic deficits (Tseng 2021)—the result of colonial approaches to bilingualism. As schools in the U.S. attend to students of minoritized language backgrounds in ways that further deficit models of bilingualism, parents and caregivers are then likely to continue the ideological process of shaming, and when that is coupled with social media technologies like TikTok, popular videos with millions of views exemplify more of what bilinguals do not do. One linguistic skill that is evident in the TikTok videos is inverted Spanglish (Rosa 2016), where, at times, Latinx Spanish-English bilinguals produce linguistic realizations that indicate intimate knowledge of both Spanish and English, such as hyperanglicizing Spanish pronunciation (e.g., sabo as [sa.boʊ] or madre as [ma.$\stackrel{⏜}{\mathrm{d}\mathsf{ʒ}}$ɹe]). In the data here, from the deficit perspective, pronunciations like these, supposed lexical gaps, and jarring repetition requests from adults may create indexes of inauthenticity. What is more, these repetition requests can take various forms, such as the example of “huh!?” in Transcript 1 and the *gasp* in Transcript 3, and both reflect linguistic hostility (Christoffersen 2019). At the same time, the linguistic productions of no sabo kids give us insight into the detailed knowledge that bilinguals have in their linguistic repertoire, and additionally, the cultures and ethnicities that are inextricable from language.

The expectations of parents, caregivers, and other listening subjects are founded on an understanding of Spanish as separate from English, and moreover, the language-related purpose of the parody videos, for example, depends on positioning one language against the other. #nosabokid commentaries and performances hinge upon and uphold colonial ideologies of named languages. This work has also drawn from colonial narratives about language separateness, as discussions of code-switching inevitably reflect the grammatical and ideological marking of a boundary between two languages (Urciuoli 2008), whereas translanguaging intentionally erases the distinction (García et al. 2021). However, if bilingual speakers are viewed as drawing from one unified network of language, as is the case within a translanguaging framework (García et al. 2021), they cannot be “deficient” in any named language. Some videos in this dataset, albeit just two of 95, demonstrate a critical call to decolonize the discussions we have around heritage Spanish speakers in the U.S. One TikToker stated in a video—“I did not choose the no sabo kid life, the no sabo kid life chose me”—and the videos that call out the negative effects of the term make explicit mention of how intra-Latinx shaming furthers linguistic assimilation to English in the U.S.

Mediated bricolage—the collective construction of a style, identity, or persona via various semiotic resources within new media—is a mechanism through which the no sabo kid persona is produced and reproduced. Regarding how sociolinguistic styles within a mediated context play critical roles in damaging and marginalizing ethnolinguistic identity, I raise two points. First, mediated instantiations of sociolinguistic styles tell us not just about the linguistic features used within a mediascape (Appadurai 1996), but also how linguistic features become enregistered through semiotic bricolage and how mediated tools (e.g., emojis) can shed light on language and its indexes of ethnicity and authenticity. Second, I highlight the negation—the “no”—in the label no sabo kids. Here, the focus is on the negative, on what the speakers do not know. In other words, the ideological stances produced by these videos create meta-communicative commentary about what U.S. Latinx bilinguals do not do, rather than the myriad of grammatical and communicative skills they dominate, in both languages.

The no sabo kid hashtag community is a mediated manifestation of ideologies surrounding U.S. Latinx bilinguals, where a supposed lack of proficiency in Spanish and grammatical blending of Spanish and English not only index inauthentic ethnicity but also point to an idea of community, namely no sabo kids. In Transcript 1, the adult listener directly asks the teenager if she is a “no sabo kid” simply due to vocalic productions that sound more like English during a cultural activity linked to latinidad. Thus, the nexus between Latinx identity and the ability to speak Spanish in the U.S. means different things according to who is speaking and who is listening, which is a result found in both mediated and non-mediated contexts (Lawrence and Clemons 2022; Rosa and Flores 2017). Furthermore, we must note how linguistic productions by young Latinx bilinguals are grammatically and cognitively valid. To productively blended verbal morphology, for example, requires high level competency in both languages (see Toribio and Bullock’s (2016) work on code-switching and proficiency) and the presence of English in discourse where Spanish is expected does not presuppose a lack of ability in Spanish.

These sociolinguistic expectations from the perspective of listening subjects (Inoue 2003) are conditioned by ideologies related to standardization (Lippi-Green 1997) and race/ethnicity (Rosa 2019). Negative attitudes towards bilingualism likely exacerbate ongoing declines of Spanish use in the home (Funk and Lopez 2022). When we consider the interconnectedness of linguistic policing and language production in light of mediatization (Androutsopoulos 2016) and how connections across digital communities are more readily available (Zappavigna 2011)—and more public—we come to understand just how prolific (i.e., millions of views) the no sabo kid persona, along with its ascribed attributes and qualities, can be. The speakers noted here are bilinguals not by choice but by necessity, and the social meanings of Spanish, at least in the context of U.S. Latinx communities on social media, are inseparable from the sociopolitical forces of English.

5. Conclusions

Latinx linguistic practices in the U.S. are policed under the colonial histories of both Spanish and English, where Latinxs are racialized to lack competency and proficiency in each language (e.g., Rosa 2019). Bilingual language use is framed as different and deficient under the guises of white monolingual norms and expectations, which are traceable back to the formation of the nation-state (Andresen and Carter 2016). Bilingualism and language contact are certainly not new phenomena, though in postmodern societies the circulation of language ideologies is more easily dispersed across boundaries and communities (Lynch 2022). One contemporary result of this has been the no sabo kid persona on social media (Stransky et al. 2022), and this article highlights the sociolinguistic mechanisms through which this persona is enregistered. The bricolage of bilingual phonology, morphology, lexical items, discursive themes, and other semiotic material, all in the context of new media, allows sociolinguists to directly consider multimodality in the construction of personae (Ilbury 2023; Mondada 2016). Additionally, this work intentionally emphasizes the interface of production and perception within third-wave sociolinguistics; the combination of how Spanish is heard (e.g., tense /i/ heard as lax /ɪ/), which body produces that variation, and the person or institution that listens, plays a significant role in creation of ethnolinguistic identities. Finally, the linguistic productions described here exist at a blurred crosslinguistic boundary between Spanish and English, yet the reactions, perceptions, ideologies are couched within monolithic understandings of speech and speech communities. It is at this juxtaposition where the devaluation of bilingualism occurs.

Future research on no sabo kids will benefit from an explicit comparison of parody versus non-parody video types to understand more completely how proficiency and competency can be understood within parody. Additionally, a quantitative approach to mediated bricolage, perhaps involving acoustic analyses, to understand how variables build upon each other in statistical terms might be enlightening, but researchers should keep in mind potential acoustic limitations of social media data. Finally, if future lines of this work are to be carried out qualitatively, it will be useful to incorporate multiple coders and test for inter-rater reliability, to avoid potential single-researcher bias.