Taikomoji kalbotyra, 23: 32–49 eISSN 2029-8935
https://www.journals.vu.lt/taikomojikalbotyra DOI: https://doi.org/10.15388/Taikalbot.2026.23.3

Digital code-switching in Lithuania: a mixed-methods study of influencers’ Instagram captions

Ovidijus Videika
Vilnius University
videika.ovidijus@gmail.com
https://ror.org/03nadee84
https://orcid.org/0009-0001-8165-0325

Birutė Ryvitytė
Vilnius University
ryvityte@gmail.com
https://ror.org/03nadee84

Abstract. Code-switching, the alternation between two or more languages within a single communicative act, has been widely investigated since the 1970s. In the 21st century, with technological advancements and globalisation, scholars have increasingly examined code-switching on social networking platforms. This paper investigates code-switching in the Instagram captions of four most-followed native Lithuanian influencers’ accounts using a mixed-methods approach, offering exploratory insights into digital code-switching in Lithuania. Quantitative analysis provided the frequencies and distributions of code-switching, the languages involved, and its types and functions, while qualitative analysis applied slightly modified classification frameworks proposed by Poplack (1980) and Halim and Maros (2014). The findings suggest that code-switching occurs frequently among the four influencers, with higher overall switching frequencies observed among the male influencers in this dataset, primarily driven by one account, and with English emerging as the dominant language. Within this sample, male influencers more frequently employ hashtag switching and the function of discoverability, while female influencers more often use inter-sentential and free switching. The functional patterns further indicate that code-switching may serve as a means for these influencers to construct different identities and position themselves within broader social and cultural contexts.
Keywords: code-switching, Lithuanian influencers, social media, Instagram, sociolinguistics

Skaitmeninė kodų kaita Lietuvoje: mišrių metodų tyrimas influencerių „Instagram“ įrašuose

Santrauka. Kodų kaita – kalbinis reiškinys, kuomet asmuo vieno komunikacinio akto metu naudoja dvi ar daugiau kalbų – plačiai tyrinėjamas nuo XX a. aštuntojo dešimtmečio. XXI amžiuje, technologijų pažangos ir globalizacijos eroje, kodų kaita socialinių tinklų platformose susilaukia vis didesnio mokslininkų susidomėjimo. Šiame straipsnyją tiriama kodų kaita keturių daugiausia sekėjų turinčių lietuvių nuomonės formuotojų „Instagram“ įrašuose, taikant mišrių metodų prieigą ir siekiant pateikti pirmines įžvalgas apie skaitmeninę kodų kaitą Lietuvoje. Kiekybinė analizė atlikta siekiant nustatyti kodų kaitos, vartojamų kalbų, tipų ir ir funkcijų dažnius bei pasiskirstymą, o kokybinė analizė atlikta remiantis šiek tiek modifikuotomis Poplack (1980) ir Halim ir Maros (2014) klasifikacijomis. Tyrimo rezultatai atskleidė, jog kodų kaita tarp keturių nuomonės formuotojų pasitaiko gana dažnai, o du vyriškos giminės nuomonės formuotojai kodus keitė dažniau nei moterys, tačiau šį skirtumą daugiausia lėmė vieno nuomonės formuotojo kalbinė praktika. Anglų kalba buvo nustatyta kaip dominuojanti kodų kaitoje. Be to, vyriškos giminės nuomonės formuotojai dažniau taiko grotažymių kaitą bei atrandamumo funkciją, o moterys – tarp-sakinio bei laisvąją kodų kaitą. Funkciniai modeliai taip pat leidžia manyti, kad kodų kaita gali būti pasitelkiama kaip priemonė konstruoti skirtingas tapatybes ir pozicionuoti save platesniuose socialiniuose ir kultūriniuose kontekstuose.
Raktažodžiai: kodų kaita, Lietuvos influenceriai, socialinės medijos, Instagram, sociolingvistika

_________

Copyright © 2026 Ovidijus Videika, Birutė Ryvitytė. Published by Vilnius University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use,
distribution, and reproduction in any medium, provided the original author and source are credited.

1. Introduction

Code-switching (henceforth referred to as CS), the practice of alternating between two or more languages within a single communicative act, has been widely studied in spoken and written contexts since about 1970s, but the rise of the digital age, and especially social media, has made the accessibility of vast amounts of data much easier. Social media provided an opportunity to easily connect with people all around the world which caused people to shift from their native language to other languages in many interactive situations (Crystal 2001: 220). Therefore, throughout the 21st century scholars have increasingly investigated CS on social networking platforms, such as Facebook, Twitter, WhatsApp, or Instagram.

1.1. Defining CS

Many scholars have attempted to define CS from different perspectives, yet the core of the concept has remained largely consistent. One of the main definitions, which dates back to 1977, was provided by Gumperz as he researched conversational CS. He defined it as a “juxtaposition of passages of speech belonging to two different grammatical systems or subsystems, within the same exchange” (Gumperz 1977: 1). Poplack (1980: 583) refined the definition by specifying that such alternation can occur “within a discourse, sentence or constituent”.

Building on these foundational definitions, subsequent scholars have expanded the scope of CS in various ways: Holmes and Wilson (2017) distinguished between CS and code-mixing; Muysken (2000) used the term code-mixing in place of intra-sentential CS; Gardner-Chloros (2009) included dialectal alternation; and Myers-Scotton (2006) extended the concept to encompass shifts in style. Furthermore, scholars have numerous times attempted to distinguish CS from borrowings claiming that the latter is grammatically and phonetically adapted to (Holmes and Wilson 2017: 44) as well as morphologically and syntactically corresponds with the target language (Poplack 2000: 205), and bears social content (Myers-Scotton 2000: 133).

Despite ongoing scholarly debate and the difficulty in drawing clear boundaries between the two, CS and borrowings in the Lithuanian context can be distinguished by following the guidelines proposed by the State Commission of the Lithuanian Language, which clearly indicate whether a particular word’s use is acceptable or should be avoided.

1.2. CS categorisations

While definitions of CS have provided a theoretical foundation, its categorisation by form and function has allowed researchers to analyse it more systematically. Poplack (1980: 589–605) was the first to categorise CS into three types according to the place in which they occur: tag-switching, inter-sentential CS (or extra-sentential CS), and intra-sentential CS. This classification has been widely adopted by scholars (Romaine 1995; Myers-Scotton 2006; Bullock and Toribio 2009; Holmes and Wilson 2017) and remains one of the most stable models, thus forming the structural basis of this study.

Subsequent scholars have proposed further refinements to the structural categorisation. Myers-Scotton further distinguished intra-word switching, where morphemes from two languages appear within a single word (1989: 343), and intra-clause switching, referring to alternation within a clause (2006: 239). In addition, Muysken (2000: 3) has expanded intra-sentential CS into three patterns: insertion, alternation and congruent lexicalisation.

Beyond structural types, functional classifications aim to explain why speakers switch. These vary by context and can be multifunctional or depend on the interpretation (Eldridge 1996: 305). For example, Eldridge (1996: 305–307) identified seven functions within the context of second language classrooms, while Appel and Muysken (2005: 118–120) outlined conversation-based functions.

However, due to similarity between research data, the present study adopts the functional classification developed by Halim and Maros (2014), who analysed CS in Facebook status updates and identified eleven distinct functions (see Section 2). Their model, which divides broader categories into more specific ones, is particularly suited to digital discourse and thus relevant to this study’s focus on Instagram captions.

1.3. Previous research on CS

As research on social media and computer-mediated communication has grown throughout the 21st century, CS in social networking contexts has attracted increasing scholarly attention. Numerous studies have explored CS across various online platforms (Androutsopoulos 2007; Eldin 2014; Al-Qaysi and Al-Emran 2017; Lubis et al. 2017; Inggitajna and Inggitajna 2021).

Facebook-based research has been particularly prominent in CS. For example, Eldin (2014) analysed Facebook status updates posted by Arabic-English bilinguals and concluded that CS functions in online written communication closely resembled those in spoken language. Similarly, Androutsopoulos (2007: 22) revealed that CS among bilingual speakers in German-based online forums mirrored conversational CS in terms of functions. Eldin (2014: 85) argued that English-Arabic bilinguals switched due to limited facility, insufficient competence, and habitual expressions, whereas Halim and Maros’ (2014: 133) study of CS in English-Malay bilinguals’ Facebook interactions demonstrated that users were competent in both languages.

Beyond Facebook, Twitter has also emerged as an important context for digital CS: Habtoor and Almutlagah (2018) examined intra-sentential CS in tweets, while Jurgens et al. (2014) discovered a significant prominence of CS among hashtags across over twenty languages, a finding broadly relevant to social media platforms including Instagram.

However, research on CS specifically on Instagram remains limited with one notable exception of Inggitajna and Inggitajna’s (2021) study which examined CS on this platform, analysing a single account and finding mostly intra-sentential switching used for message qualification. While informative, the study’s narrow scope highlights the need for broader, corpus-based research into how Instagram users employ CS.

1.4. Previous research on CS in Lithuania

In Lithuania, CS has been studied across various contexts, including TV commercials (Vaicekauskienė and Šmitaitė 2010), conversations among Vilnius teenagers (Vyšniauskienė 2012), Facebook (Jakelienė 2018; Miliun 2020) and digital communication across various platforms (Darginavičienė and Ignotaitė 2020). Studies focusing on social media have primarily focused on Facebook, with Jakelienė (2018: 21) identifying English as the dominant language for CS in posts of public figures, serving functions such as discourse marking, emphasising, attracting attention and establishing identity and Miliun (2020: 113) discovering that CS frequently functioned as a directive, with users switching languages based on their interlocutor.

A broader perspective was provided by Darginavičienė and Ignotaitė (2020) who examined CS across several platforms, including Instagram, YouTube, Facebook, and Twitter. However, due to a small and varied sample, the authors provided only brief generalisations, suggesting that CS was often used to enhance the visual appeal of digital texts (Darginavičienė and Ignotaitė 2020: 412). As a result, it did not yield substantial conclusions about CS on any specific platform or within any defined user group. Taken together, these studies demonstrate growing scholarly interest in CS on social media in Lithuania but underscore the need for platform-specific, large-scale investigations.

1.5. Aims and research questions of the current study

While CS has been widely analysed across different modes of communication, including on social media, most studies to date have focused primarily on Facebook and Twitter, leaving Instagram comparatively underexplored. To address this gap, the present study investigates instances of CS in Instagram captions posted by native Lithuanian influencers. By employing quantitative and qualitative methods, this article aims to answer the following research questions: 1) how frequently the selected influencers code-switch on Instagram; 2) whether male or female influencers code-switch more often within the selected sample; 3) which languages, types, and functions of CS are employed by the selected influencers and how their frequencies differ by gender; and 4) what sociolinguistic observations can be drawn from the functional analysis of CS across the selected influencers.

2. Data and methods

For this study, an influencer is defined as a person who shapes their audience’s attitudes, opinions, and behaviour by means of blogs, status updates, photo captions, and the use of other social media (Freberg et al. 2011: 90). The data were collected from Lithuanian influencers’ Instagram photo captions which offer a wide range of CS cases suitable for the analysis. The influencers were selected based on the follower count at the time of data collection; therefore, the two most followed women, Karolina Meschino (F1) and Agnė Jagelavičiūtė (F2), and the two most followed men, Rolandas Mackevičius (M1) and Naglis Bierancas (M2), were chosen. The dataset comprises 634 captions posted between January 1 and June 30, 2021. While based on a limited sample of four influencers and 634 captions, the study offers exploratory rather than generalisable insights into digital CS practices.

Each caption from the selected time period was manually extracted and analysed. All captions were grouped into four subcorpora by author and then compiled into two main corpora based on gender. No alterations were made to grammar, punctuation, capitalisation, symbols, or emojis in order to preserve the original meaning. While Instagram is an inherently multimodal platform, a full multimodal analysis was beyond the scope of this study. The analysis therefore focuses primarily on captions, which constitute a linguistically stable and replicable unit of analysis. Multimodal affordances, such as images and emojis, were taken into account where relevant for interpretation but were not analysed systematically. For clarity, each caption in the corpus was coded using an abbreviation that combines the influencer’s gender (F or M), a number distinguishing between the two influencers of each gender (1 or 2), and a sequential entry number (e.g., F1-33 refers to the 33rd caption in the first female influencer’s subcorpus). F1 subcorpus comprises 3,139 words, F2 – 6,777, M1 – 1,633, and M2 – 2,605.

Captions written solely in the Lithuanian language were excluded from the analysis. Therefore, any manifestation of another language—excluding borrowings, loanwords, and neologisms that are approved by the State Commission of the Lithuanian Language—was treated as a case of CS. Similarly, proper nouns, specifically names of products, brands, and companies, that include other languages, were not treated as CS cases as they are non-translatable words.

All posts containing CS were annotated, with foreign-language elements marked and categorised by type and function. As the study aims to present both numerical and categorical findings of CS, both quantitative and qualitative methods were employed. Quantitative analysis was applied to estimate frequencies of CS languages, types, and functions, with normalisation per 1,000 words to account for subcorpus size differences. In addition to descriptive statistics, chi-square tests of independence were performed to explore potential associations between gender and selected variables (languages, structural types, and functions). To maintain statistical validity, infrequent categories (e.g., rarely used languages or functions) were merged into broader “Other” groups to avoid low expected frequencies. Given the small number of selected influencers, these tests were interpreted as indicative rather than inferential.

Qualitative analysis was conducted by applying CS categorisation into types identified by Poplack (1980) as well as functional classification designed by Halim and Maros (2014). The former has presented three types of CS: tag-switching, inter-sentential, and intra-sentential (Poplack 1980: 589–605) while the latter has provided a CS classification into eleven functions: switching for quotations, addressee specifications, reiterations, message qualifications, clarifications, emphasis, checking, indicating emotions, availability, the principle of economy, and free switching (Halim and Maros 2014: 129–132). All instances of CS were manually annotated according to each classification by the first author and inter-coder reliability was not tested, which constitutes a limitation of the study. Nevertheless, other consistent criteria were applied throughout.

Additionally, hashtags were separated into an individual type and function as they occur on their own (not in the context of a sentence) and have their fixed function on the Instagram platform. According to Chang and Iyer (2012: 248), hashtags are “a unique tagging format” connecting posts to user-construed concepts, which are primarily used in order to group the post with other people’s posts that include the same hashtag. Therefore, the post appears “within global discussion” or becomes “a part of a virtual community” (Jurgens et al. 2014: 52). For this reason, it was necessary to modify the adopted classifications and establish a separate type of hashtag switching as well as a function entitled switching for discoverability.

3. Results and discussion

In the following sub-sections, the research findings are presented, covering the frequency and distribution of CS, the languages involved, its types, and functions. The functions sub-section is further divided into individual categories, each illustrated with examples from the data to highlight the most common CS occurrences.

3.1. Frequency and distribution of CS

CS was identified in 122 captions in F1, 114 in F2, 35 in M1, and 86 in M2 subcorpora. Since multiple switches may occur within a single caption, the total number of CS instances reached 871: 238 from F1, 263 from F2, 63 from M1, and 307 from M2. On average, this amounts to 61.5 CS cases per 1,000 words across all captions.

Table 1. Overall raw and normalised frequencies of CS cases in Instagram captions of Lithuanian influencers individually and according to gender

As shown in Table 1, male influencers in the dataset exhibit higher overall code-switching frequencies than female ones; however, this pattern is largely driven by M2, whose extensive use of code-switching substantially increases the aggregated male frequencies. M2 thus emerges as the most frequent code-switcher in the dataset, with 117.9 CS cases per 1,000 words. The second highest frequency is observed for F1, who uses approximately 75.8 foreign words per 1,000 in her captions. F2 and M1’s captions contain roughly the same number of CS cases per 1,000 words. The data also indicate that female influencers tend to write longer captions overall.

(1) Jau gyvenu atostogų mood’e <…> Cant wait!!!
#winterwonderland <…> (M2-26)
‘I’m already in the mood for vacation <…> Can’t wait!!!
#winterwonderland <…>’

Example (1) from M2 illustrates a typical caption containing several CS instances in a single post, including intra-sentential (mood’e), inter-sentential (Can’t wait) and hashtag switching (#winterwonderland, #homesweethome, etc.), as well as three different functions, free switching for the first CS case, indicating emotions for the second one, and discoverability for the third. Therefore, this example also demonstrates how different types and functions can co-occur within a single caption.

3.2. Languages used for CS

CS cases that were found in Lithuanian influencers’ photo captions were firstly analysed by identifying their languages. The results have shown that the analysed influencers used a wide variety of foreign languages for CS. In total, 13 different languages were used across influencers’ captions, including English, Italian, Russian, and others.

Table 2. Raw (RF) and normalised frequencies (NF) of thirteen identified CS languages in Instagram captions of Lithuanian influencers, individually and according to gender

As Table 2 reveals, English was used most frequently by both male and female influencers which aligns with broader national trends showing increased English proficiency among younger Lithuanians, who increasingly treat it as a global lingua franca. Statistical data collected in Lithuania in 2021 shows that the proportion of English-speaking residents has nearly doubled since 2001, rising from 16.9% to 31.1% (Statistics Lithuania 2021).

(2) Pilnas pokalbis Gyciu Ivanausku jau Online. (M1-12)
‘Full conversation with Gytis Ivanauskas is now online.’

Example (2) demonstrates a typical use of English in captions, especially for technology- and social media-related vocabulary. In this particular case, M1 is using an intra-sentential English word online which means “available or done on the internet” (Cambridge Dictionary 2025) and which does not have an accurate equivalent in Lithuanian.

Surprisingly, all influencers used Italian at least once, but females used it more frequently in comparison to males. This can be explained through the demographic information of influencer F1, who is bilingual with her father being Italian.

(3) Grazie zia @magagninfrancesca per il stupendo vestitino della piccola Isa <…> (F1-30)
‘Thanks, aunt @magagninfrancesca for the beautiful little dress for baby Isa <…>’

Example (3) illustrates how F1 is using the Italian language. The analysis revealed that most of her Italian CS cases are inter-sentential and very rarely appear as other types. In this case, F1 is addressing an Italian person which requires a switch to this language.

What stands out in Table 2 is M2’s extensive usage of the Russian language. This language mostly appeared as abbreviations of Russian swearwords (zjbs, px, nx, pzda, blet, etc.) which are used as expletives in informal contexts to convey certain emotions towards situations or people. On the other hand, F2 used Russian to quote a popular saying narubyt na nasu which means ‘remember this once and for all’ demonstrating her Russian language proficiency.

A chi-square test of independence revealed a statistically significant (χ²(3, N = 871) = 72.63, p < .001) association between gender and the three most used languages for CS (English, Russian, Italian), with all remaining low-frequency languages combined as “Other”. Male influencers used Russian significantly more frequently, while female influencers were more likely to employ English and Italian, indicating both linguistic backgrounds (as in the case of the bilingual F1) and stylistic preferences (as in the case of Russian swearwords).

Other languages occurred less frequently, from one to three times per influencer, often linked to cultural references or personal contexts. For instance, Hebrew chebra (‘squad’, informal) and its variations, Arabic-derived kaifas (‘pleasure’, informal) and Persian-influenced bazaras (‘conversation’, informal) were used in casual, informal speech since these words belong to Lithuanian slang. Other languages were used while travelling (Spanish), as references to food (Serbian, Turkish), to quote songs (Portuguese) or phrases (Latin) and to express emotions (Polish and French).

3.3. Types of CS

The classification of CS types in this study followed the model provided by Poplack (1980), which includes intra-sentential, inter-sentential, and tag switching, with the addition of hashtag switching as a fourth category to reflect the Instagram-specific context.

Table 3. Raw (RF) and normalised frequencies (NF) of four CS types in Instagram captions of Lithuanian influencers individually and according to gender

As shown in Table 3, the results among influencers vary considerably. F1 primarily engages in inter-sentential switching, producing full sentences in a foreign language, while M1 more frequently employs intra-sentential switching by inserting foreign words or phrases into otherwise Lithuanian sentences. The following examples illustrate their usage of inter-sentential and intra-sentential switching.

(4) Oggi avresti compiuto 90 anni... <…> Nonna, quanto ci manchi... (F1-7)
‘Today you would have turned 90... <…> Grandma, how much we miss you...’

(5) <…> bandau išlaužti jogos pozą, nuo kažko gi reik pradėt? Nes iki sixpacko toli (M1-10)
‘<…> trying to master a yoga pose, you need to start somewhere? Because I’m a long way from getting a six-pack’

As can be seen from example (4), F1 is using inter-sentential switching in the form of whole Italian sentences. This case was also assigned the function of addressee specification as F1 is addressing her Italian grandmother, even though she has passed away but would have turned 90 that day. Example (5), on the other hand, demonstrates M1’s intra-sentential CS as he embeds the English word six-pack and also morphologically adapts the word by adding the Lithuanian genitive case ending -o.

M2 stands out for his exclusive use of hashtag switching, with 74.5 cases per 1,000 words, far more than any other influencer. Consider example (6).

(6) <…> #loveislove #gaylife <…> #zjbys <…> (M2-77)

Here, M2 uses hashtags in English (love is love, gay life) and Russian (zjbys—a Russian swearword used in Lithuanian slang to express strong positive emotions, roughly equivalent to ‘awesome’). These hashtags likely serve a dual purpose: aligning the post with broader online conversations and increasing its visibility on the platform.

The results, as indicated in Table 3, have also shown that Lithuanian influencers do not use tag switching at all except for F2. However, in comparison to other types, tag switching in her captions appears relatively seldom.

(7) <…> Padovanok man dar viena diena ir man pakaks.
Pradziai, i mean. <…> (F2-81)
<…> Just give me one more day, and that will be enough.
For starters, I mean. <…>’

From example (7), it can be seen that F2 uses I mean which serves as a typical tag switch. This phrase is frequently used in informal contexts by speakers who want to clarify their previous statement. Notably, F2 adheres to the punctuation norms by separating the phrase with a comma.

Overall, the findings reveal that, within this dataset, female influencers more frequently favour inter-sentential switching, whereas male influencers more often use hashtags and intra-sentential switching. The relationship between gender and structural types of CS was also found to be statistically significant (χ²(3, N = 871) = 93.44, p < .001), confirming that female influencers are more likely to produce full sentences in a foreign language while males embed foreign words or phrases within Lithuanian structures or use them in hashtags.

3.4. Functions of CS

To determine and explore CS functions in the captions of Lithuanian Instagram influencers, this study employed the functional classification proposed by Halim and Maros (2014), with results summarised in Table 4.

Table 4. Raw (RF) and normalised frequencies (NF) of twelve CS functions in Instagram captions of Lithuanian influencers individually and according to gender

As presented in Table 4, F2 and M2 mostly switch for discoverability, which reinforces the idea that hashtags function to increase post visibility by targeting a broader international audience. In contrast, F1 and M1 primarily use free switching, which suggests stylistic motivations or a desire to demonstrate foreign language proficiency. For example, F1 often writes in Italian, a language likely unfamiliar to most of her Lithuanian audience, potentially using it to showcase her competence or cultural affiliation.

The findings also reveal that F1 frequently uses CS for addressee specifications, likely due to her participation in public events such as award ceremonies and TV shows where she acknowledges stylists, designers and make-up artists by crediting them in English. On the other hand, F2 and M2’s captions often function as free switching, suggesting that CS may also serve as a trendy or expressive choice in the current digital landscape.

Strikingly, none of the influencers code-switched for reiteration or clarification. According to Halim and Maros (2014: 130), switching for reiteration involves restating information in another language to enhance clarity, while switching for clarification involves elaborating on a point in a different language. The absence of these functions might demonstrate that the analysed influencers appear to orient toward an audience presumed to be proficient in Lithuanian and English; therefore, they do not need to repeat or clarify themselves in Lithuanian when they use an English word or phrase and vice versa. However, the use of foreign languages other than Lithuanian and English may reduce accessibility or lead to misinterpretations, especially if users rely on Instagram’s often-inaccurate built-in translation tools.

Figure 1. Distribution of CS functions (per 1,000 words) across two genders

As Figure 1 demonstrates, male influencers tend to code-switch for discoverability substantially more, while female influencers most frequently use free switching. These findings may suggest that, within this dataset, male influencers employ CS to expand their audience reach internationally, whereas female influencers use it more for stylistic purposes and to display their foreign language skills.

Female influencers employed switching for quotations and addressee specifications more frequently than males, while all other functions were more common among males. However, functions such as quotations, message qualifications, emphasis, checking, and availability appeared at relatively similar frequencies between genders.

To verify these tendencies, a chi-square test was applied to examine the association between gender and four functions that are observed to be significant in the descriptive statistics (discoverability, free switching, indicating emotions and addressee specifications), with all remaining low-frequency functions grouped under “Other”. The result was statistically significant, χ²(4, N = 871) = 137.79, p < .001, confirming the above interpretations.

The least used functions by both female and male influencers were switching for message qualifications and checking. This may indicate that the selected influencers assume their audience to be predominantly Lithuanian; therefore, they rarely present a topic in one language and then elaborate on or question it in another.

The following sub-sections will focus on each function individually as well as provide examples from Lithuanian influencers’ captions along with explanations and interpretations.

3.4.1. Switching for quotations

According to Halim and Maros (2014: 129), CS might be used in order to quote speeches of other people or oneself. The following example illustrates a typical instance of this function, in which song lyrics are quoted.

(8) I’m not calling for a second chance
I’m screaming at the top of my voice
<…> (F2-5)

As seen in example (8), F2 employs inter-sentential CS by quoting the lyrics from the English-language song “Same Mistake” by James Blunt (2007). The caption appears beneath a photo of F2 and her partner, and the song’s theme (reflecting on past relationship failures and vowing to avoid repeating them) suggests a deliberate and meaningful choice.

Overall, quotation-based CS was relatively infrequent across the dataset. This function was mostly realised through English song lyrics and widely known online phrases, with occasional quotations from films, magazines or public figures. Such instances of CS reflect the influencers’ tendency to align with global popular culture, thereby reinforcing a relatable and up-to-date online identity. In the Lithuanian digital space, the use of English quotations positions these influencers as modern and culturally attuned figures who participate in a globalised media discourse.

3.4.2. Switching for addressee specifications

Halim and Maros (2014: 130) argued that this strategy is used to directly address others. In their analysis of Facebook interactions, the authors also highlight the platform’s tagging function, which allows users to mention other individuals by name, thereby directing content toward them (Halim and Maros 2014: 130). Instagram offers a similar feature, enabling users to tag other accounts by including the @ symbol followed by a username. In the present study, Lithuanian influencers were found to frequently employ CS for addressee specifications when tagging others in their captions.

(9) Me and my mom, on vacation in Dubai, in bathroom watching breaking news about @navalny. As loud as possible.
<…> (F2-2)

Example (9) demonstrates F2’s use of inter-sentential CS when addressing another person—in this case, Alexei Navalny, a well-known Russian opposition leader. The caption was posted shortly after his arrest in Russia, and the switch to English, paired with the tag @navalny, suggests a politically conscious use of language intended to express solidarity and support.

Switching for addressee specifications appeared quite frequently across the dataset, particularly in captions produced by female influencers. In most cases, switching occurred when addressing a person and combining this with Instagram’s tagging function. However, a few instances involved addressees that were not individuals or accounts, and thus tagging was not employed. This function most often appeared when giving credit to professionals the influencers collaborated with, including photographers, make-up artists, hair and fashion stylists, as well as when tagging foreign individuals or companies. Interestingly, some tagged accounts referred to Lithuanian people or brands, yet CS was used due to the broader thematic or linguistic context of the caption. The use of this function reflects the analysed influencers’ orientation toward both local and international audiences and seems to serve as a tool for constructing a networked and cosmopolitan identity. By integrating CS within professional tagging, influencers seem to index their collaborative and participatory practices within digital culture.

3.4.3. Switching for message qualifications

Switching for message qualifications occurs when a speaker introduces a topic in one language and then elaborates or expands on it in another (Halim and Maros 2014: 130). As shown in the quantitative data, this function was used only by F1 and M2. In both cases, the influencers began their captions with an English sentence, followed by further elaboration in Lithuanian. Consider example (10).

(10) It’s unbelievable what a woman’s body can do! Laikau savo lėlytę ant rankų ir negaliu suprast niekaip dar kad čia mano vaikutis <…> (F1-32)
‘It’s unbelievable what a woman’s body can do! I am holding my little girl in my hands and still can’t believe that this is my baby <…>’

In example (10), F1 introduces the topic of childbirth in English and proceeds in Lithuanian to express her awe and emotional reaction. The same introductory phrase in English is later reiterated in Lithuanian; however, it was not classified as switching for reiteration since the English statement precedes the Lithuanian elaboration and not vice versa.

Other instances of message qualifications appeared in captions covering topics such as owning a bathtub for the first time, advertising sex toys, travelling and having a photoshoot. In all cases, CS occurred at the beginning of the caption, with English being the only language used to introduce the topic before moving into elaboration in Lithuanian. Overall, this function was relatively rare and limited to two influencers, yet the identified patterns suggest that English serves as an attention-grabbing device, signalling the influencers’ orientation toward both international and national audiences. Moreover, this function illustrates how a foreign language can be employed to construct a dual persona that alternates between global and local identities, international accessibility and cultural rootedness.

3.4.4. Switching for emphasis

Halim and Maros (2014: 130) explained this function as switching to another language to stress or draw attention to specific information within a statement. In the present study, most cases of CS assigned to this function also featured capitalisation, an orthographic strategy that further reinforces emphasis.

(11) IT’S @bykarolinameschino BIKINI SEASON <…> (F1-27)

In example (11), F1 switches to the English language and uses capital letters to maximise emphasis. It is also evident that the switch serves to attract attention of not only Lithuanians but also worldwide audiences, as she promotes her own fashion brand. At the time of the post, a new bikini collection had been launched, and the caption seeks to highlight this release with marked urgency and excitement.

Switching for emphasis appeared predominantly in captions promoting giveaways, advertisements of products, or exciting announcements. All instances involved the use of English, and most were also combined with capital letters to create a sense of immediacy and excitement. The use of this function serves as a marketing strategy, while the choice of English indexes linguistic prestige and alignment with global branding discourse. Through such usage, influencers build an aspirational identity that follows international lifestyle trends and participates in a global consumer culture.

3.4.5. Switching for checking

According to Halim and Maros (2014: 131), switching for checking is used when a person seeks reassurance, validation, recommendations, or opinions from others. This function takes the form of tag questions, yes/no questions or wh-questions (Halim and Maros 2014: 131). In the present study, all instances of this function appeared in interrogative form, as illustrated in example (12) below, aligning with Halim and Maros’ observation.

(12) Mondays are for fresh starts? ar labiau nx? <…> (M2-18)
‘Mondays are for fresh starts? or more like to hell with it? <…>’

As can be seen, example (12) features a multilingual instance of switching for checking. M2 uses an inter-sentential switch to English for the first part of the question and an intra-sentential switch to Russian in the second part. The abbreviation nx, derived from a Russian profanity, is often used to convey strong negative emotions towards something or someone and has become common in Lithuanian slang. By juxtaposing two contrasting views of Mondays, M2 is inviting his followers to respond with their own perspectives.

Although switching for checking was relatively infrequent in the dataset, the observed instances appeared either as rhetorical questions, which did not require direct responses from the audience but served to provoke thought, or as conversational ones inviting engagement and opinions. The example above was the only case that included a language other than English, while all remaining instances employed English exclusively. The use of English once again facilitates accessibility and enables a wider audience reach, reflecting its role as a digital lingua franca that allows these influencers to construct inclusivity and interact with their international audiences.

3.4.6. Switching for indicating emotions

According to Halim and Maros (2014: 131), people may switch codes to express emotions and feelings, as some sentiments might be conveyed more effectively by using other languages. The present dataset revealed that this function most often involved short, emotionally charged expressions that add emphasis or convey affective stance, as in example (13) below.

(13) Einam gatve, Lukas stabteli ir klausia: Ar nori,kad prie šito pastato nufotkinčiau? Kaip zjbys, kad žino be žodžiu <…> (M2-30)
‘We’re walking down the street, Lukas stops and asks: ‘Do you want me to take your photo by this building?’ How awesome is it that he knows without me saying a word <…>’

Example (13) reveals yet another instance of a Lithuanian slang term derived from a Russian expletive. M2 uses an abbreviation zjbys which originally is vulgar in Russian; however, it is often used colloquially among Lithuanians to express strong positive emotions and can be loosely translated as ‘awesome’ or ‘really cool’. The emotive function is further reinforced by the inclusion of emoticons, a pattern Halim and Maros (2014: 131) also noted in their study.

Switching for indicating emotions was moderately frequent, with most cases appearing in both English and Russian languages across the dataset. English was typically used to express positive or enthusiastic emotions, including such instances as omg, chill, excited and lol, while Russian-derived slang tended to convey stronger affective intensity and informality, as illustrated in example (13). The use of emotive CS demonstrates how the analysed influencers draw from different linguistic inventories to achieve varied emotional effects—English highlights a cosmopolitan persona, while Russian signals local familiarity and relatability.

3.4.7. Switching for availability

This function, according to Halim and Maros (2014: 131), is used when finding an accurate translation in the native language is more difficult than expressing the same term in its original language, or when translation might cause semantic inaccuracy. In the present dataset, such cases typically involved specific cultural or lexical items for which Lithuanian equivalents either do not exist or are not widely known among users.

(14) Receptas keturiems asmenims:
-500g Fusilloni <…>
-300g ajvar <…> (F1-1)
‘Recipe for four people:
-500g of Fusilloni <…>
-300g of ajvar <…>’

Example (14) shows F1 code-switching while sharing her cousin’s pasta recipe. This caption alone contained seven Italian and two Serbian terms, including fusilloni (Italian for a spiral-shaped pasta), burrata and parmigiano reggiano (both Italian for types of cheese) as well as ajvar (Serbian for a relish made from bell peppers). F1 may have used the original names to avoid inaccurate Lithuanian translations and to help followers find the products in grocery shops.

Switching for availability was relatively infrequent but consistent across the dataset, encompassing references to food as exemplified above, travel destinations such as Holbox, Napoli and Zakynthos, and Instagram-specific terminology without exact Lithuanian equivalents such as reels (short videos) and highlight (a platform function for preserving stories on the user’s profile). The use of such terms prevents semantic inaccuracy while simultaneously indexing cultural and platform-specific affiliation. In this way, CS serves not only as a strategy for referential accuracy but also as a marker of the analysed influencers’ intercultural identity and digital literacy.

3.4.8. Switching for the principle of economy

This strategy is adopted when a person uses foreign words that are less complex than their native equivalents (Halim and Maros 2014: 132). For instance, a person might switch to English because the word is shorter or has fewer syllables than the equivalent of that word in the native language, making communication easier (Halim and Maros 2014: 132).

(15) Kolegos insomniakai ir siaip ivairus beprociai, <…> (F2-43)
‘Colleagues who are insomniacs and various other crazy people, <…>’

In example (15), F2 uses morphologically adapted English word insomniac to describe people with sleeping problems. The Lithuanian equivalent, “nemigos kankinamas žmogus”, is considerably longer and more complex to write, illustrating the use of a foreign word to make communication and caption writing more economical.

Switching for the principle of economy appeared moderately across the dataset, with most cases involving English words and phrases that are considerably shorter than their Lithuanian equivalents, such as Wi-Fi (Lith. ‘belaidis ryšys’), brand (Lith. ‘prekės ženklas’) or besties (Lith. ‘geriausi draugai’). These switches simplify expression and enhance fluency within the digital space, where speed and immediacy are highly valued. Such use reflects how global media norms shape Lithuanian language practices and indexes influencers’ social identity aligned with digital culture. Moreover, influencers demonstrate their linguistic proficiency through the spontaneous selection of foreign words for efficiency and ease of use.

3.4.9. Free switching

According to Halim and Maros (2014: 132), free switching occurs when a CS case cannot be assigned to any other function and there is no apparent reason for the switch. This function also includes instances where CS serves a stylistic purpose or demonstrates competence in another language (Halim and Maros 2014: 132). In this study, CS cases involving words used in Lithuanian slang (e.g., chebra, bazaras) or social media-related language (for example, share, tag, comment, like, scroll, etc.) that have direct, widely-known Lithuanian equivalents.

(16) <…> Kai sumigdau vaikus, prisedu paskrolinti Soulz.lt online shope <…> (M1-18)
‘<…> When I put the kids to sleep, I sit down to scroll through Soulz.lt online shop <…>’

As can be seen in example (16), M1 uses English social media terms scroll and online shop, morphologically adapted with Lithuanian affixes (prefix pa-), inflections (infinitive verb form), phonetics (letter c is pronounced as [t͡s] while k as [k]) and cases (locative case inflected by the ending -e). This illustrates the integration of foreign lexicon into Lithuanian with morphological adjustment, reflecting the use of CS to fill lexical gaps or for stylistic effect.

Overall, free switching was relatively frequent across the dataset, emerging as the most used function by analysed female influencers, and often reflected linguistic creativity and playfulness. By integrating English social media terms, slang expressions and other foreign words into Lithuanian discourse, the influencers demonstrate both multilingual competence and stylistic awareness. Such instances of CS index an urban and youthful identity, where language mixing is used as a marker of modernity and digital fluency. This function allows influencers to express individuality and align themselves with cosmopolitan online communities in the Lithuanian digital space.

3.4.10. Switching for discoverability

The analysis revealed that most of the hashtags used by Lithuanian influencers were cases of CS, as they primarily contained languages other than Lithuanian. It was also found that these hashtags typically correspond with the caption or photo content.

(17) <…> Ledai greit pasirodys prekyboje, būtinai paragaukit (mano meilė kur su peanut butter) <…> #zeroaddedsugar #omg #icrecream <…> #zerosugar (M2-4)
‘The ice cream will be in stores soon — definitely try it (my favorite is the one with peanut butter) <…> #zeroaddedsugar #omg #icrecream <…> #zerosugar’

In (17), M2 uses English hashtags in a sponsored post for ice cream. Therefore, in order for hashtags to correspond with the post as well as to reach a wider audience, he uses the English language and phrases that match the product’s attributes (#icecream, #zeroaddedsugar and #zerosugar) as well as his personal emotions towards the product (#omg).

Switching for discoverability was the most frequent function among selected male influencers, appearing approximately 3.5 times more often than among females. The majority of hashtags in the dataset contained foreign-language elements, most commonly English, with one instance having 19 out of 20 hashtags in English. Such density illustrates how English operates as a global lingua franca within digital spaces such as Instagram, optimising post visibility through platform algorithms. By employing English and other foreign languages, selected influencers enhance the discoverability of their content among international audiences and simultaneously construct an influencer identity which is shaped by the affordances of the platform and its inherently global reach.

4. Conclusions

This study aimed to quantitatively and qualitatively examine CS among four Lithuanian influencers on Instagram, identifying the languages, types, and functions used. The findings show frequent use of code-switching across the examined Instagram captions, with English functioning as the primary switching language alongside twelve additional languages. Although aggregated frequencies are higher in male-authored captions, this pattern is largely shaped by the extensive use of foreign-language hashtags by a single male influencer rather than by a consistent gender-based tendency. Accordingly, hashtag switching functioning for discoverability is particularly prominent in male-authored captions. Female influencers within this sample, on the other hand, most frequently employ inter-sentential CS and the function of free switching. The functional analysis further demonstrated that the studied influencers use CS not only as a communication method but also as a tool to construct cosmopolitan, urban and intercultural identities and signal global orientation, digital fluency, networking and affiliation with both local and international audiences. Overall, digital CS seems to emerge as a strategic practice shaping how influencers present themselves socially, emotionally and culturally within the Lithuanian digital landscape.

Due to the limited scope of this study, captions from only four influencers (two male, two female) were examined, with limited variation in demographic characteristics. This means that the findings allow to draw only tentative conclusions about digital CS trends in Lithuanian influencers’ Instagram profiles. The reliance on manual coding without inter-coder reliability testing further narrows the generalisability of the results, thereby making them exploratory rather than conclusive. Nevertheless, this study provides a solid foundation for further investigation of digital CS among Lithuanian influencers.

Future research could examine a larger and more diverse sample of influencers, incorporating additional demographic variables such as age, bilingualism, and sexual orientation. Furthermore, a multimodal analysis of CS—including visual (photos, videos, emojis), auditory (music, sounds), and platform-specific features (location tags, other account tags)—could yield a deeper understanding of how CS operates in social media contexts, especially on such a multifaceted platform as Instagram.

Author’s Contribution

Ovidijus Videika: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Resources; Validation; Visualization; Writing – original draft; Writing – review & editing.

Birutė Ryvitytė: Conceptualization; Methodology; Resources; Supervision; Writing – review & editing.

Abbreviations

CS Code-switching (when used as a noun)

F1 First female influencer

F2 Second female influencer

M1 First male influencer

M2 Second male influencer

NF Normalised frequency

RF Raw frequency

Data sources

Instagram profile @agne_stiliusos. Available at: https://www.instagram.com/agne_stiliusos (data collected between 1 January and 30 June 2021).

Instagram profile @karolinameschino. Available at: https://www.instagram.com/karolinameschino (data collected between 1 January and 30 June 2021).

Instagram profile @naglimantas. Available at: https://www.instagram.com/naglimantas (data collected between 1 January and 30 June 2021).

Instagram profile @rolandas. Available at: https://www.instagram.com/rolandas (data collected between 1 January and 30 June 2021).

References

Al-Qaysi, Noor & Mostafa Al-Emran. 2017. Code-switching usage in social media: A case study from Oman. International journal of information technology and language studies, 1(1), 25–38. https://www.researchgate.net/publication/318982373_Code-switching_Usage_in_Social_Media_A_Case_Study_from_Oman (accessed 15 August 2025).

Androutsopoulos, Jannis. 2007. Language choice and code switching in German-based diasporic web forums. In Brenda Danet & Susan C. Herring (eds.), The multilingual internet: Language, culture, and communication online, 340–361. New York: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780195304794.003.0015.

Appel, René & Pieter Muysken. 2005. Language contact and bilingualism. Amsterdam: Amsterdam University Press.

Blunt, J. 2007. Same mistake [song]. On All the lost souls. Atlantic; Custard.

Bullock, Barbara E. & Almeida Jacqueline Toribio. 2009. The Cambridge handbook of linguistic code-switching. Cambridge: Cambridge University Press.

Cambridge Dictionary. 2025. https://dictionary.cambridge.org/ (accessed 1 July 2025).

Chang, Hsia-Ching & Hemalata Iyer. 2012. Trends in Twitter hashtag applications: Design features for value-added dimensions to future library catalogues. Library trends, 61(1), 248–258. https://doi.org/10.1353/LIB.2012.0024.

Crystal, David. 2001. Language and the internet. Cambridge: Cambridge University Press.

Darginavičienė, Irena & Indrė Ignotaitė. 2020. Code-switching in the computer-mediated communication. RUDN journal of sociology, 20(2), 405–415. https://doi.org/10.22363/2313-2272-2020-20-2-405-415.

Eldin, Ahmad Abdel Tawwab Sharaf. 2014. Socio linguistic study of code switching of the Arabic language speakers on social networking. International journal of English linguistics, 4(6), 78–86. https://doi.org/10.5539/ijel.v4n6p78.

Eldridge, John. 1996. Code-switching in Turkish secondary school. ELT journal, 50(4), 303–311. https://doi.org/10.1093/elt/50.4.303.

Freberg, Karen, Kristin Graham, Karen McGaughey & Laura A. Freberg. 2011. Who are the social media influencers? A study of public perceptions of personality. Public relations review, 37, 90–92. https://doi.org/10.1016/j.pubrev.2010.11.001.

Gardner-Chloros, Penelope. 2009. Code-switching. Cambridge: Cambridge University Press.

Gumperz, John Joseph. 1977. The sociolinguistic significance of conversational code-switching. RELC journal, 8(2), 1–34. https://doi.org/10.1177/003368827700800201.

Habtoor, Hussein Ali & Ghzail Faleh Almutlagah. 2018. Intra-sentential code-switching among bilingual Saudis on Twitter. International journal of linguistics, 10(2), 1–18. https://doi.org/10.5296/ijl.v10i2.12915.

Halim, Nur Syazwani & Marlyna Maros. 2014. The functions of code-switching in Facebook interactions. Procedia – social and behavioral sciences, 118, 126–133. https://doi.org/10.1016/j.sbspro.2014.02.017.

Holmes, Janet & Nick Wilson. 2017. An introduction to sociolinguistics, 5th edn. London/New York: Routledge.

Inggitajna, Clara & Annisa Agra Inggitajna. 2021. The use of code switching in sunnydahye’s Instagram caption. UC journal: ELT, linguistics and literature journal, 2(1), 14–21. http://dx.doi.org/10.24071/uc.v2i1.3244.

Jakelienė, Eglė. 2018. Code-switching on Facebook in Denmark and Lithuania. Taikomoji kalbotyra, 10, 1–25. https://doi.org/10.15388/TK.2018.17441.

Jurgens, David, Stefan Dimitrov & Derek Ruths. 2014. Twitter users #CodeSwitch hashtags! #MoltoImportante #wow. In Mona Diab, Julia Hirschberg, Pascale Fung & Thamar Solorio (eds.), Proceedings of the first workshop on computational approaches to code switching, 51–61. Doha: Association for Computational Linguistics. https://doi.org/10.3115/v1/W14-3906.

Lubis, Indah Sari, Satyawati Surya & Adinda Usin Muka. 2017. The use of code-switching among the late adolescents in social media Facebook. CaLLs, 3(2), 83–95. http://dx.doi.org/10.30872/calls.v3i2.817.

Miliun, Violeta. 2020. Kodų kaita skirtingų lyčių asmenų feisbuko paskyrose. Šalčininkų rajono atvejis. Taikomoji kalbotyra, 14, 99–115. https://doi.org/10.15388/Taikalbot.2020.14.8.

Muysken, Pieter. 2000. Bilingual speech: A typology of code-mixing. Cambridge: Cambridge University Press.

Myers-Scotton, Carol. 1989. Codeswitching with English: Types of switching, types of communities. World Englishes, 8(3), 333–346. https://doi.org/10.1111/j.1467-971X.1989.tb00673.x.

Myers-Scotton, Carol. 2000. Code-switching as indexical of social negotiations. In Li Wei (ed.), The bilingualism reader, 127–153. London: Routledge. https://doi.org/10.1515/9783110849615.151.

Myers-Scotton, Carol. 2006. Multiple voices: An introduction to bilingualism. Malden: Wiley-Blackwell.

Poplack, Shana. 1980. Sometimes I’ll start a sentence in Spanish y termino en español: toward a typology of code-switching. Linguistics, 18(7/8), 561–618. http://dx.doi.org/10.1515/ling.1980.18.7-8.581.

Poplack, Shana. 2000. Sometimes I’ll start a sentence in Spanish y termino en español: toward a typology of code-switching. In Li Wei (ed.), The bilingualism reader, 205–240. London: Routledge. https://doi.org/10.4324/9780203461341-24.

Romaine, Suzanne. 1995. Bilingualism, 2nd edn. Oxford: Wiley-Blackwell.

Statistics Lithuania. 2021. Results of the 2021 Population and Housing Census of the Republic of Lithuania. https://osp.stat.gov.lt/2021-gyventoju-ir-bustu-surasymo-rezultatai/pratarme (accessed 26 July 2025).

Vaicekauskienė, Loreta & Reda Šmitaitė. 2010. Anglų kalbos vartojimas ir kodų kaita Lietuvos televizijų reklamose. Kalbotyra, 62(3), 108–125. https://doi.org/10.15388/Klbt.2010.7647.

Vyšniauskienė, Inga. 2012. Polilingvali Vilniaus jaunimo raiška: socialinės tapatybės paieškos. Kalbotyra, 64(3), 140–157. https://doi.org/10.15388/Klbt.2012.7665.

Ovidijus Videika is an MA student of English Studies (Media Discourse) at the Faculty of Philology, Vilnius University, Vilnius, Lithuania, and works as an Editor at Notified. His research interests include sociolinguistics, corpus linguistics, critical discourse analysis and multimodality. He is currently working on his MA thesis investigating online hate speech discourse.

Dr Birutė Ryvitytė, now retired, was an Associate Professor at the Faculty of Philology, Vilnius University, Vilnius, Lithuania. Her research interests include linguistic pragmatics, discourse analysis, corpus linguistics, sociolinguistics, contrastive rhetoric, LSP theory and language teaching for specific purposes. During her academic career, she taught courses in pragmatics, sociolinguistics, research methods, business English and rhetoric, among others.

Submitted September 2025

Accepted January 2026