Gender Differences in Influencer Advertisements on Instagram: A Multimodal Perspective

Congruent with the rapid development of internet users worldwide, social media advertising has been mushrooming on various social network platforms, including Instagram. Among the many forms of advertising that marketers have relied on is influencer advertising, which pertains to hiring popular Instagram influencers to endorse particular products or services on their Instagram accounts. This study explores how these Instagram influencers use both texts and images to convey their advertising messages. Further, the study explores the differences in how male and female influencers construct their endorsement posts. A total of 20 advertisement posts were collected from 10 influencers and were analyzed from the multimodal approach integrated with the concepts of linguistic metafunctions in Systemic Functional Linguistics. Then, the gender differences in texts and images were elaborated. It is shown that interpersonal language dominates the language used in the captions, despite the specific differences in the linguistic items used by men and women. Likewise, in the images, the influencers seem to consistently put the spotlight on themselves rather than on the products to captivate the followers' attention. The findings of this study will eventually contribute to the literature on advertising language and multimodal analysis.


Introduction
Advertising is undeniably one of the most common texts found across different media. Through a visit to public places such as malls or an internet search, one can instantly come across various advertisements on a wide range of topics. It is argued that advertising has then become one powerful marketing tool to attract non-consumers to become consumers of their product or service (Jaworska, 2020). In other words, an effective advertisement should be persuasive enough to lead the viewers to purchase the products or service eventually.
comprising an advertisement are not so straightforward as its purpose. The presence of discursive and semiotic resources-primarily texts and images-renders the analysis of the language used in advertisements relevant to consider.
Advertising has developed prodigiously due to the advancement of technology and communication tools. The previous advertising trends in the pre-internet era were mainly centered around print advertisements found in newspapers, magazines, billboards, and posters, but with the advent of the internet, web advertising or online advertising has become equally essential to marketers. The potential of online advertising is signified by the fact that more and more people worldwide are considered online active users. According to a survey conducted by Hootsuite and We are Social (2021), up to 4.80 billion people, accounting for around 60.9% of the world population, have been connected to the internet. Of the whole internet users, a staggering total of 4.48 people have already used social media, accentuating the potential of advertising through social media. Social media advertising has been confirmed to effectively convince the viewers to purchase the advertised products or services (Dehghani & Tumer, 2015;Gaber et al., 2019). Among the many available social media platforms, Instagram stands out as one social media with an enormous number of users. To the advertising potential, Instagram is said to attract a total of 1.39 billion potential viewers. Therefore, it is essential to consider the discourse found in these Instagram advertisement posts.
Instagram per se has a wide range of advertisement forms, but one unique method of advertising that has been mushrooming is influencer advertisement. Influencer marketing is an effort in which celebrities are paid to upload a promotional post of the products or services on their media channels (Lee & Kim, 2020). The use of influencer marketing on social media, in conjunction with the proliferation of internet users across the globe, has garnered considerable interest among advertisers. Influencer Marketing Hub (Geyser, 2021) revealed that the growth of influencer marketing was estimated at around 9.7 billion US dollars in 2020 and is expected to reach around 13.8 billion dollars in 2021. Empirical research has likewise attested to the effectiveness of influencer advertisements (Hermanda et al., 2019;Lee & Kim, 2020;Nam & Dân, 2018;Soni, 2016;Wielki, 2020). Social media influencers can improve the brands' image and thus influence the consumers' purchasing decisions positively (Hermanda et al., 2019;Lou & Yuan, 2019). As well as that, the Instagram users typically view these influencers as credible reviewers, as they have had a sense of trust in the influencers before they can trust the products (Nam & Dân, 2018). Considering the significance of influencer advertisement, the ways these influencers construct their advertising messages in their Instagram posts to portray the products become worth scrutinizing.
In light of the importance of social media advertising discourse, linguistic research has hitherto focused on the discourse of both printed advertisements (Ananda et al., 2019;Fuertes-Olivera et al., 2001;Koteyko, 2015;Lirola & Chovanec, 2012) and online advertisements (Hidarto, 2021;Labrador et al., 2014;Luqyana & Imperiani, 2021;Zjakic et al., 2017). Some of these studies have revealed how advertising language exhibits several idiosyncratic features characterizing how advertising messages should be conveyed. For instance, Koteyko (2015) conducted a corpus study and revealed six linguistic categories of lexicogrammatical features typically found in printed advertisements. Similarly, Labrador et al. (2014) analyzed web-based advertisements based on a small corpus of online ads collected and enumerated the standard linguistic features advertisers typically use. There has been a strong emphasis on the role of text in representing the persuasive nature of advertisements.
However, many advertisements depend on images in conveying the advertising messages, and with that said, the roles of images should not be overlooked in the advertisement analysis. At this point, one growing approach towards linguistic analysis may offer some contribution: the multimodal approach.
As argued by Jewitt et al. (2016), the multimodal approach is considered the appropriate approach to analyzing texts Journal of Language and Literature ISSN: 1410-5691 (print);2580-5878 (online) Anderson Hidarto & Aryani Andrieza consisting of different semiotic modes, namely texts and images. This is because the interaction between the two semiotic resources may evoke a new meaning. Kress (2010) also asserted that mere analysis of the language in texts with multimodal nature might obscure the actual meaning conveyed in the texts. Since Instagram is a photo-based social media (Newberry, 2020), the multimodal approach is deemed the most relevant analysis model. Regarding the adoption of this approach towards online advertisement analysis, it remains relatively underexplored. There were some studies examining the roles of images in printed advertisements (Ananda et al., 2019;Lirola & Chovanec, 2012;Nugroho, 2009), but with Instagram posts, only one such study was conducted by Hidarto (2021), who focused on influencer posts on Instagram and revealed how both texts and images indeed played a role in strengthening the persuasive messages of their product endorsement posts. Therefore, this study uses the multimodal approach to ensure a complete picture of the roles of different modes in the influencers' Instagram posts.
In addition to employing the multimodal perspective, the present study explores gender differences in how males and females construct their endorsement posts on Instagram. The differences have been observed in many previous studies, the most notable of which is the study by Newman et al. (2008), who enumerated the linguistic features typically used by either males or females. They collected an exhaustive database of over 14,000 texts files, which they analyzed to shed light on whether, how, and why men and women used linguistic items differently. In terms of the lexical and grammatical items, men used more articles, quantifiers, spatial words, and most notably swearing words, contrary to women who used more personal pronouns, intensive adverbs, and emotion words. In addition, Newman et al. also pinpointed how males and females tended to lean towards different themes in their language use, with the former using impersonal and object-focused language (e.g., occupation, politics, sports) and the latter using more people-oriented language (e.g., family, social life). The insights from Newman et al. (2008) attest to the notion that males and females use language differently in some ways. Furthermore, the use of distinct language features by both genders can also be observed in their social media posts, including those on Facebook (Park et al., 2016), Twitter (Hu & Kearney, 2020), and Instagram (Yuliani et al., 2020). Park et al. (2016) pinpointed the differences in topics associated with both genders: female Facebook users tended to talk more about social-related topics (e.g., friends and family), whereas the male users focused more on discussing objects or giving arguments, further complemented with swearing and expression of anger. In the case of Instagram, Yuliani et al. (2020) observed similar findings in that women were more likely to share personal stories, in contrast to men who shared less personal content. They also revealed that females used more implicit language features (e.g., hedging expressions) than males who relied on direct and explicit language (e.g., swear words).
The analysis of how both males and females use images to express themselves has also been conducted by some earlier researchers (Döring et al., 2016;Kapidzic & Herring, 2015), who have found several differences in the way the users portray themselves in self-portraits or self-photographs (selfie). Likewise, gender differences are also argued to exist in the discourse of advertising (Goffman, 1988;Bui, 2021), but many studies have merely concentrated on the representation, roles, and stereotypes of both genders in advertisements (An & Kim, 2006;Huhmann & Limbu, 2016;Plakoyiannaki et al., 2008;Sari & Ganiem, 2021), overlooking the actual ways both men and women use both texts and images in expressing their persuasive messages. Hence, it is intriguing to explore how male and female influencers construct their Instagram advertisements, namely the pictures and captions, and identify the differences in the language and image use between the two genders.
Considering the research gaps above, this study aims to explore the gender differences in the language use of Instagram influencers in their advertisement posts. The following research questions guide this study:

1.
How do male and female influencers use texts to convey their advertising messages on Instagram? 2.
How do male and female influencers use images to convey their advertising messages on Instagram? 3.
How do the texts and images interact in the advertisements of both male and female influencers?

Data Collection
The data consisted of texts and photos from 20 advertisements posts uploaded by ten social media influencers on Instagram-five males and five females. Most of the selected posts were uploaded in 2021 to ensure the recentness of the Instagram posts. Regarding the credibility of the selected Instagram accounts, only those with blue verification badges which provide public access were chosen. Following Instagram's website's terms of use and community guidelines, a verification badge allows people to know that the accounts they are following are real people.
The themes of the chosen advertisements were fashion products, with two posts taken as a sample from each influencer. In choosing the Instagram posts, it is essential that the photos were followed by some captions with enough text to analyze. Additionally, English should be used as the dominant language in the captions, but captions using another language such as Indonesian were still considered. The selected posts were those posted in 2021 to ensure the novelty of the posts. Another criterion was that the photos should portray the influencers instead of only showcasing the products, which was essential to the image analysis of how the influencers-both males and females-portrayed themselves.

Data Analysis
There were several steps to multimodal analysis carried out in the present study. To address the first research question on the role of texts in the Instagram posts, the lexicogrammatical elements of the texts were analyzed and classified based on the three types of linguistic metafunctions elaborated by Halliday and Matthiessen (2014) in Systemic Functional Linguistics (SFL): ideational, interpersonal and textual metafunctions. The ideational metafunction refers to the use of language to convey information about a topic, for example, modifications with adjectives (e.g., a stylish, comfortable jacket) or quantifying expressions (e.g., a variety of, a wide range of). The interpersonal metafunction refers to the use of language to establish a relationship between the speaker and the interlocutor, including pronouns (e.g., you, we), past tenses, or opinion markers (e.g., in my opinion, I think). Textual metafunction refers to the structure of the posts (e.g., titles, headings) in portraying the meaning and the use of connectives (e.g., first, most importantly, plus). The numbers of occurrences of items from each metafunction were then calculated and compared for both genders. One caveat was that one clause might contain more than one linguistic metafunction, meaning that the features in such a clause can be classified into different metafunctions. For instance, the clause 'But never forget to always wear my comfortable sandals everywhere' contains the three metafunctions, namely ideational (i.e., comfortable), interpersonal (i.e., my), and textual (i.e., but). Finally, A chisquare analysis was performed to test whether gender differences may significantly lead to the use of different linguistic metafunctions. A significance value of less than 0.05 (p < .05) indicates that genders may not necessarily result in the reliance on different metafunctions in the language use of both males and females.
As to the more detailed gender analysis, the present study also used the perspectives from Newman et al. (2008), who listed down the linguistic features usually used by males or females. After the occurrences of the lexicogrammatical items were counted, the initial analysis focused on the features that both males and females used equally frequently, as indicated by the similar number of occurrences. The identification of gender differences was then based on the lexicogrammatical items that females used significantly more frequently than males, and Journal of Language and Literature ISSN: 1410-5691 (print); 2580-5878 (online) Anderson Hidarto & Aryani Andrieza vice versa. These items would account for how males and females used language differently in their Instagram captions.
To address the second research question, the image analysis was carried out adopting the approach of Visual Grammar by Kress and Van Leuween (2006). The visual composition of the photographs of each influencer was analyzed, consisting of three elements: information value, salience/prominence, and framing. Information value refers to the visual positioning of the elements in the images (top or bottom, left or right, center or margin), salience is related to weights of all visual elements, which are subject to size, foregrounding, contrasting colors, and sharpness and framing is the presence (or absence) or frames to maximize the connection (or disconnection) between elements in the image.
The gender differences in the images were then identified using the perspectives from Goffman (1988) and Kang (1997), who listed the variables to analyze self-portraits. Goffman (1988) pinpointed several variables, such as relative size (e.g., women are shown smaller or lower compared to men), feminine touch (e.g., women tend to touch themselves), function ranking (e.g., occupation), the ritualization of subordination (e.g., women tend to be laying down), and licensed withdrawal (e.g., women are used as background in the picture). Kang (1997) further added more criteria, namely body display, and self-assertiveness. Body display refers to how the female body is portrayed (e.g., body-revealing clothes or nudity), while self-assertiveness is related to the critical cues shown by the person in the image (e.g., the use of hands, facial expressions, body positioning). In case of any emerging elements, these elements would be added accordingly into the analysis.
Finally, to address the third research question, the interaction between the texts and images was scrutinized using the Integrative Multi-Semiotic Model (Fei, 2004). At this point, it was interesting to examine what roles the images played concerning the texts. These roles may be in the form of assertion or complementation (Lahiani, 2004). The assertion function, or homospatiality (Fei, 2004), refers to how both semiotic modes operate similarly to strengthen each other's message. For example, the image may prove what is mentioned in the text and vice versa. Meanwhile, complementation, or semiotic metaphor (Fei, 2004), is observed when the presence of one semiotic mode may change the meaning of the other. To illustrate, texts may be interpreted differently with or without the images, meaning that the latter serve to complement the meaning of the texts.

Results and Discussion
The findings of the study are comprised of three major sections. The first section focuses on exploring the gender differences in the Instagram captions. The second elaborates the gender differences in the visual cues (i.e., the Instagram photos). Finally, the last section scrutinizes the relation between the captions and images in conveying the persuasive messages in the advertising posts.

Gender Differences in the Texts
After all the 20 advertisements are analyzed, each of the discrete linguistic items can be categorized into three linguistic metafunctions. The overall distribution of the occurrences in the advertisement posts of both males and females is shown in Table 1. Furthermore, to test the significance of the influence of genders on the use of linguistic metafunctions in the Instagram posts, the chisquare analysis is performed with a significance value of .05, as shown in Table 2.  As indicated in Table 1, female influencers used more persuasive features in their posts than their male counterparts. What is similar is that both groups primarily relied on interpersonal language use in their endorsement posts (56.7%), followed by the ideational and textual metafunctions (28.3% and 15.1%, respectively). From the percentages, female influencers used more ideational and textual linguistic features, whereas male influencers used more interpersonal features. However, the chisquare analysis in Table 2 demonstrated how gender differences did not significantly influence the use of the three linguistic metafunctions (χ 2 = 5.004; α = .082, p > .05), meaning that being male or female did not necessarily lead to the tendency to use one type of metafunction in one's advertisement posts.
The prevalent use of interpersonal language features in the influencers' posts aligns with Hidarto's (2021) findings, which demonstrated the dominance of colloquial items in the posts regardless of gender. The insignificant influence of gender on the choice of the three metafunctions is attributable to the notion that research on gender and language should take into account the contexts wherein the language is used (Gu, 2013). At this juncture, the choice to use more interpersonal language is emblematic of the nature of the context of the influencer advertisement, that is, Instagram.
Instagram provides a platform for influencers to engage with the audience. As suggested by Haenlein et al. (2020), what is more important in influencer marketing is to establish relationships with their followers to gain trust and credibility. As a consequence of such trustworthiness, the followers' perceptions towards the advertised brands can be more positive, and they are more likely to purchase the products (Hermanda et al., 2019; Lou & Yuan, 2019). This clarifies the tendency for influencers-whether males or females-to look and sound friendly to the audience, rather than explaining too many details about the products.
Despite the insignificant difference at the level of general distribution, it is essential to scrutinize the differences in the use of specific lexicogrammatical items, specifically in each of the three linguistic metafunctions.

Ideational Metafunction
The ideational metafunction is typically aimed to convey information. In advertisements, it provides more information about the products being advertised. Hidarto (2021) and Labrador et al. (2014) suggested that any lexicogrammatical items used to evaluate the promoted products positively can be classified into this language use. Table 3 below shows the distribution of linguistic items denoting ideational use. As shown in Table 3, there are some similarities and differences in the captions posted by both male and female influencers. The first similarity is evident in clauses, notably main clauses and verbless clauses similarly. Both gender groups tend to use a complete sentence in their caption, for instance, 'I have been looking for a pair of Ghurka Sandals, and was super excited when I was introduced to @fortuna_shoes_indonesia a few weeks ago.' Not only that, but some influencers also omitted the verb altogether, producing verbless sentences as in 'Now available online through katespade.co.id.' Another identical feature used by both males and females is the insertion of testimonies about the brand or products (e.g., This is perfect for the current weather, @colehaan.id always impresses me with its comfort and style) and the narration of personal stories (e.g., Ready to get through the day with the coolest collab between #CoachxMBJ!).
The prevalence of personal testimonies and stories is inextricable to the nature of Instagram as a personal blog (Ramos-Serrano & Martínez-García, 2016). By narrating personal stories, the influencers try to maintain the impression of their Instagram as a personal journal, wherein they are willing to demonstrate their personal characters in the interaction with the followers. As Kusyanti et al. (2019) have posited, Instagram can serve as a platform for self-representation. As the sole owners of the Instagram accounts, the influencers have the right to display a persona they intend to show to their followers, independent of the corporate restrictions typically found in companies' blogs.
Nevertheless, under further scrutiny, the way male and female influencers shared their product testimonies or stories differ in some ways. Females are observed to add more descriptions about the advertised products, using more descriptive adjectives (e.g., a spacious bag, luxurious and comfy mood), intensifying adverbs (e.g., supernatural, subtly luxurious, it makes me look even cuter), and prepositional phrase (e.g., with a versatile look, by @aden_hijab, from @everleigh.studio) than the males. Additionally, the topics found in the females' posts are mainly peopleoriented (e.g., Grab bags that is now one of my new partners, my future human baby) or emotion-related (e.g., we love a casual but cute ootd, super excited, I let out all my love). On the other hand, male influencers focus more on less emotive object-oriented topics to refer to the advertised products (e.g., so the fabric uses environmentally safe process, use less water, clothes recycling). The more frequent use of descriptive, emotion-related, and people-oriented words or expressions among female influencers conforms to previous studies (Gultom et al., 2020;Newman et al., 2008;Park et al., 2016;Wahyuningsih, 2018). As revealed by Park et al. (2016), such phenomenon was pertinent to the tendency for women to emphasize "psychological and social processes" (p. 2). Meanwhile, men were portrayed to focus more on objective and impersonal topics, that is, those focusing on non-animate objects. Park et al. related such inclination to the idea that women are likely to show warmth while men tend to be 'colder'. In the case of Instagram captions, the present study also supports the results from Gultom et al. (2020), who found that women were inclined to be more expressive than men in their Instagram posts.

Interpersonal Metafunction
The interpersonal metafunction refers to how the influencers use the words to engage with their followers, establishing a relationship. What can be categorized into this linguistic use includes items denoting inclusivity (e.g., the pronoun you or we, imperatives) and those characterizing colloquial or informal language use (e.g., reduction, language play, use of first-person pronouns). Table 4 enumerates the specific language items in the interpersonal metafunction. As presented in Table 4, both men and women use informal reduced grammatical items, including subject mission (e.g., is just me being a tourist, really love it) and clipping (e.g., cause, pict, 'em, fave). The use of punctuations, especially contraction and exclamation marks, is evident for both genders (e.g., they'll love to help you!, don't miss it!, the super dope look!). In line with the prevalence of narratives, these influencers tend to use present tenses, rather than past tenses, with first-person singular pronouns (e.g., I also add a pair of these adorable earrings). Finally, both influencer groups also use emojis and hashtags to represent emotions (e.g., smiley face) or objects (e.g., shoe emoji) instead of words.
The use of many informal-and often ungrammatical-linguistic features is emblematic of how informality is a natural consequence of online communication (McCulloch, 2019). The use of reductions, for instance, represents the need for efficiency in online communication. Similarly, emojis on Instagram replace the need to convey facial expressions or emotions usually present in face-to-face communication. Besides that, as Gultom et al. (2020) noted, by using more colloquial words, the influencers attempt to establish a personal relationship with the followers, known as synthetic personalization (Fairclough, 2010). Fairclough further explains how such a strategy can be associated with promotional discourse, as in advertisements.
Besides similarities, there are also some noticeable differences. Female influencers use more second-person pronouns as a direct way to address their followers (e.g., They'll love to help you, discounts code just for you), unlike the males who use more imperative words (e.g., check them out, tap @kanmo for more!). Females are also observed to use more hedging phrases (e.g., Just like this sandal) and language play like alliteration (e.g., every edelweiss will be everlasting).
The difference in linguistic features can result from how influencers build intimacy with their followers. The use of second-person pronouns among female influencers in contrast to the use of imperatives among males might indicate the indirect and direct nature of talk for both genders, respectively (Park et al., 2016;Yuliani et al., 2020). Female influencers possibly want the followers to construe the post as an 'invitation' to use the products, whereas the men consider it 'directives' for their viewers. Additionally, indirectness is possibly associated with tentativeness (Newman et al., 2008), indicating some degree of politeness instead of aggression. The language play present in the females' posts also indicates their attempt to exhibit uniqueness, a factor contributing to establishing their credibility as opinion leaders (Casaló et al., 2020). The males tend to retain their nature of 'objectivity,' focusing on adding information rather than attitude words.

Textual Metafunction
The textual metafunction is concerned with how the text is organized to convey the messages. This metafunction includes any linguistic items used to add more ideas or those used as deictic markers. The details of the textual items are displayed in Table 5. Journal of Language and Literature Vol. 22 No. 1 -April 2022ISSN: 1410; 2580-5878 (online) As presented in Table 5, both males and females use almost all textual metafunction elements in their captions. These influencers use pronouns similarly in their Instagram caption; for example, head to their IG account, check out this bag, they will deliver it straight to your front door. Other than similarities, there are also differences which these influencers used in their text, such as different types of conjunctions with female influencers are found to use temporal conjunctions (e.g., I now realize that simply being alive is a gift) and causal conjunctions (e.g., If you're still holding on, keep staying alive ok!) more than the males. Meanwhile, male influencers use more additive conjunctions in their captions (e.g., Anyway I recently got my new line of jewelries, and they are all fantastic, I also paired it with the citysole sneakers).
The use of more additive conjunctions among male users is related to their tendency to share informative details about the products rather than sharing more details about why or how they use the products. The female counterparts are observed to use a wider variety of conjunctions to complement their narratives about the products or their lives (Newman et al., 2008).

Gender Differences in the Images
The influencers' images were analyzed based on the tenets of Visual Grammar (Kress & Van Leuween, 2006), namely the information value, salience, and framing. Afterward, the self-portrayal of each influencer is scrutinized using the variables enumerated by Goffman (1988) and Kang (1997) to identify any gender differences further.

Visual Grammar Analysis
The first compositional aspect is information value, which refers to the arrangement of information within the visual space, including left-right, top-bottom, and center-margin positions. Regarding this element, the images of both male and female influencers are observed to be exhibit some similarities. Kress and van Leuween (2006) noted that what is put on the left tends to be the information that the viewers may already know, whereas anything on the right is considered 'new' information that is emphasized in the image. In the influencers' posts, the context of the image (e.g., the place, theme) is put first on the left, followed by the influencers wearing the advertised clothing items, which means that the focus is on the influencers and the products themselves. For example, as shown in Figure 1, the setting of the image, represented by stairs, is a homey place, which corresponds to the clothing item being advertised, that is, pajamas. Similarly, the focus on the top of the image is the influencer's face, representing the 'ideal.' This implies that the influencers, including the one shown in Figure 1, intend to establish familiarity with the viewers by showing their faces that the followers surely already know. The rest of the picture downwards represents the 'real' information, referring to what the influencer advertises. Finally, in almost all the images, the products, including clothes, bags, shoes, are situated in the center of the image, becoming the main highlight of the post, as depicted in Figure 1, where the pajamas are exactly in the center of the dimension.

Figure 1. Female Influencer 'CO'
The second element is salience, which pertains to highlighting an image, represented by visual features such as size ratio, lighting, foregrounding, shot angle, and shade. One general observation of the images of most influencers is that the influencer is always the primary focus of the image rather than the product. Some samples are observed in Figure  2. In Figure 2, it is noticeable how the size of the influencer compared to the product (i.e., the sandals) occupies more space with a full shot from head to toes, despite the sandals being the product advertised. Simultaneously, to ensure that the products can still captivate the attention of the followers, they are situated in the foreground of the image in the same frame as the influencer. In a similar way, the influencer in Figure 3 takes up more visual space than the endorsed shoes, with the photo taken from a full-body angle, yet the shoes are still in the foreground of the photo.

Figure 4. Male Influencer 'CJ'
In a few posts, the male influencers may shift the prominence of the image to the products, as illustrated in Figure 4. The shoes are situated in the image's foreground, with the influencer in the background. However, regarding the shot angle, the whole body of the influencer is still shot, indicating the importance of influencers to show themselves.
Lastly, framing is related to how the elements in an image are placed together as one unit (i.e., maximum connection) or discrete units (i.e., maximum disconnection) through frame boundaries or color schemes. In most of the influencers' photos, the primary elements, that is, the influencers, are situated as a unit integrated with the background.

Figure 6. Male Influencer 'RH'
As displayed in Figure 5, the background shows a room with a typical Japanese wooden structure, added by the sunlight in the bottom frame. Through the color scheme, it is noticeable that the color of the clothes matches the natural wood, and the color of the pants is in line with the light exposure in the bottom frame. The products are appropriate to the context set by the photo's background. On the other hand, Figure 6 shows a clear boundary between the influencer and the wall behind him, with a contrasting color of black and white. The cat is another separate element from the influencer with its light color.
All the images show that the influencers, rather than the products, seem to be the main highlight of the images, with the products often serving as a complement. This is attributable to the role of the influencers as opinion leaders who can shape the followers' perceptions about the products and eventually influence their purchasing decision (Casaló et al., 2020). The followers are likely to be more familiar with the influencers rather than the brands, so the presence of the influencers serves as a catalyst for the viewers to look at the posts and heed the products. In addition, by displaying themselves, the influencers try to show originality in their posts, meaning that they want the followers to really trust the idea that they themselves use the products. The power of the interpersonal relationship among Instagram users is where influencer marketing has more potential than typical corporate marketing.

Gender Differences in Self Portrayals
It has been shown how the visual compositions demonstrate many identical features irrespective of the influencers' genders. However, the research shows that more gender-linked variations can be found in the way influencers portray themselves, as enumerated in Table 6. The only similarity between the male and female influencers is the postures that they show in their photos, including standing and sitting, as shown in Figure 7. This does not confirm the findings from Döring et al. (2016), who found that more men were standing while more women were sitting when portraying themselves. However, as we can see in Figures  7 and 8, many other variables confirm the stereotypes of how females portray themselves in photos.

Figure 8. Comparison between Influencers 'AS' and 'TT'
The other variables show some distinctions in both genders. As illustrated in Figures 7 and 8, male influencers tend not to show the standard features associated with feminine traits, such as imbalance (e.g., tilting bodies or head, feminine touch, licensed withdrawal (e.g., loss of control and withdrawal of gaze), and sparse clothing. On the other hand, all these features are present in the females' photos. These results corroborate Goffman's (1988) and Kang's (1997) variables of feminine gestures among females.
Interestingly, one notable feature emerges in the data for both males and females. As seen in Figure 8, male influencers may show a more expansive gestural space with elbows away from the body sides, which is a sign of masculinity (Prescott, 2017),

Text-Image Relation
In addition to the separate analysis of the texts and images, it is noteworthy to scrutinize the relationship between the two modes. Following Fei's (2004) concept of the Integrative Multi-Semiotic Model, there are relatively equal numbers of cases of either homospatiality (assertion), as presented in Figure  9, and semiotic metaphor (complementation), as presented in Figure 10.
In Figure 9, the image serves as the 'given' information to show what the product looks like and in what setting the product is appropriate to be worn, in addition to the influencer herself. The text then 'asserts' this information by providing the brand name of the products and how to purchase them. In comparison, Figure 10 shows an example of complementation, where the image and the text are not seemingly related but add meaning to each other. As illustrated, the caption does not mention anything about the product and narrates the expression of the influencer, whereas the image shows the brand's name on the clothing. The analysis shows how both types of text-image relations are evident among male and female influencers, entailing no difference in the advertisements posted by both genders.

Conclusion
The findings demonstrate how female and male influencers use Instagram to promote certain products. Overall, it is evident that interpersonal language is the most commonly used feature compared to ideational and textual features, although there is no significant influence of gender on the use of the three linguistic functions. However, further analysis reveals how men and women may display some differences in how they use certain words or display themselves in the images. The study has then added to the literature on gender-linked language in advertising.
The research should be perused with some caveats in mind. One notable limitation is that the influencers' motives behind using certain words or images have yet to be explored, and thus, future researchers can fill the gap by conducting interviews with Instagram users. Another caveat is the effectiveness of the multimodal posts in the view of the followers. Future studies may focus on what the followers perceive from the endorsement posts, for example, what they focus on upon seeing the posts. Finally, the present study explores one type of advertisement, fashion advertisements. Different products may entail different ways of conveying messages, which may be of further interest.
Nonetheless, the findings of this study will contribute to the scholarly research of discourse analysis, notably from the multimodal approach. This study will add more insights into how multimodal concepts can be implemented to analyze online advertisements, particularly Instagram posts that use multiple semiotic resources at once. Further, this study can serve as one definitive study of how multimodal analysis can be applied in analyzing online advertisements concerning gender influence. On the practical side, the study can benefit advertisers in that they are more aware of how online advertisements can appeal to the audience through both texts and images. Drawing on the present findings, marketers can use multiple semiotic modes to strengthen the persuasive messages in their social media ads. Language teachers can also benefit from the present study in that they can start integrating multimodal texts into their language classes,