The (Lack of) Salience of T/V Pronouns in Professional Communication: Evidence from an Experimental Study for Belgian Dutch

Abstract

In their quest to find a suitable tone of voice in an informalizing society, many companies are confronted with the choice of using T or V pronouns in their B2C communications. This paper presents an experimental study addressing the question of whether the recipients of these messages actually notice the difference between being addressed with a T form, which carries social meanings of informality and proximity, or a more distant V form, and to what extent the presence of additional informal linguistic features influences the salience of a pronoun switch. We furthermore investigate to what extent the professional socialization of participants impacts on the noticing of pronoun use. In a case study for Belgian Dutch, participants (N = 279) were presented with two versions of an information letter that they were asked to read quickly. The texts were manipulated for the use of T/V pronouns, as well as, depending on the condition, a number of additional informal linguistic features (i.e., informal punctuation, intensifiers, and English lexical items). Participants were not warned in advance about the changes between the two versions of the stimulus text. In a salience test following the presentation of the two text versions, less than 10% of participants noticed a switch in T/V form regardless of the presence of additional informal features. Similarly low rates of noticing were found for the other informal features, except for English loanwords. No differences were found depending on whether participants had a language-related professional background (e.g., language teachers, journalists, editors). We argue that the lack of noticing T/V pronouns may be due to the specifics of the Belgian Dutch system of pronominal address that has an additional highly salient colloquial pronoun of address which may obscure the difference in social meaning between the standard T and V pronouns. The discussion critically evaluates the implications of the study for the use of T/V pronouns in professional communication, musing on the complex relationship between noticing and evaluating.

1. Introduction

This paper aims to empirically investigate to what extent participants with and without a language-related professional background notice changes in T/V pronouns in B2C communication (i.e., business-to-customer communication) and how this compares to other markers of informal discourse. As argued below, many companies are in search of a more informal brand identity that presents them as accessible and customer-oriented, and in that context wonder how best to address their clients in professional communications. In T/V languages, this raises the question of which second-person pronoun is most suitable. In this paper, we focus on one such language, Belgian Dutch, which offers an interesting case given the sometimes-heated public debate on T/V pronoun use. While there is hence quite some research that reports on the use of Belgian Dutch T/V pronouns in general, little is known about how T/V differences are perceived by readers, particularly in professional communications. In our study, we set out to feed this debate on T/V usage in Belgian Dutch by answering the question of whether readers actually notice differences in use between the two forms and whether they are more salient than other markers of formality. We also aim to find out whether a change in pronoun use is more salient to readers with a professional language-oriented eye. In doing so, we contribute a perceptual perspective to both the academic study of pronoun use in Belgian Dutch and the applied question of how to communicate most effectively in professional contexts, as well as the societal debate of which pronouns suit this type of communication. To this end, we first provide the necessary background information on language as a tool to construct corporate identity (Section 2.1), on T/V pronouns in (Belgian) Dutch (Section 2.2), and on noticing and salience (Section 2.3). Building on this, we formulate precise research questions (Section 3). The paper then outlines the methodology of the study (Section 4), and reports (Section 5) and discusses (Section 6) the results.

2. Background

2.1. Corporate Identity and Language in B2C Communication

Building a strong corporate identity is crucial for companies to secure their place in an ever more competitive market (Balmer and Gray 2003). That identity is expressed via visual means through a graphic identity, but also language is a powerful tool to build a company’s image, as is well documented in research on corporate tone of voice, brand linguistics, and marketing more broadly (e.g., Usunier and Shaner 2002; Delin 2005; Kelleher 2009; Carnevale et al. 2017; Holmqvist et al. 2017). This possibility to use language to shape identity is hardly surprising from a sociolinguistic perspective: it has been studied extensively how language users build their identity by drawing on socially meaningful language variants, viz. linguistic forms that are associated with social information about the speaker or context (e.g., Eckert 2008; Eckert 2012). A well-studied example is the alveolar pronunciation of word final –ing [ɪη] as –in [ɪn] in English which has for instance been shown to carry associations with a speaker’s regional background and perceived intelligence (e.g., Campbell-Kibler 2007, 2009, 2010). Similarly, companies can deploy socially meaningful language variation to manipulate and communicate their corporate identity, which is an increasingly pressing need in today’s changing requirements for business communications.

Traditionally, corporate communication has been considered to comprise formal genres in which considerable distance exists between the sender of a message and the receiver (Campbell and Roberts 2007). However, in the context of the late modern informalization of society, many businesses have followed the New Work Order (Gee et al. 1996; Roberts 2011), a term introduced to describe the shift from the traditional ‘old’ economy and its hierarchical and formal workplace to a ‘new’ economy with its (seemingly) more egalitarian and informal structures (see also Hull 2001). Resultingly, companies often wish to build a more informal image and communication style (Norrby and Hajek 2011; House and Kádár 2020; De Dijn and Van De Mieroop 2021), which calls into question the required formality level of professional interactions and correspondence (Scheuer 2001). This search for an appropriate degree of formality in tone of voice recurs frequently in questions addressed to professional copywriters and it is these bottom-up questions from the field of B2C communication that triggered the research reported in this paper. More specifically, recurring requests for advice directed to the fourth author of this paper, who is professionally active in copywriting, revealed that companies struggle to decide on how to address customers in their communications.1 A question that was asked particularly frequently in that respect is the choice of address pronoun to use in languages that maintain a T/V distinction (Brown and Gilman 1960): is the use of more informal T pronouns an option in professional communication, or is it better to stick to more formal V forms? While there seem to be many questions and insecurities on this topic on the side of the sender, one can wonder how salient the use of T/V pronouns actually is for the recipient of a message. Hence, the aim of the present study is precisely to find out whether readers notice the use of T rather than V pronouns in corporate communications, and how the salience of T pronouns interacts with other linguistic elements representing an informal tone of voice. As a case study, we focus on T/V variation in Belgian Dutch B2C information letters (see Section 4.2), given the prominent position in societal metalinguistic debate of pronominal variation in the Belgian context (Van De Mieroop et al. 2016).

2.2. T/V Pronouns in (Belgian) Dutch

Dutch is a pluricentric language, viz. a language with more than one national centre (Muhr 2016), which is mainly used in the Netherlands and Belgium. In Belgium, it is spoken in Flanders, the northern half of the country. Both countries have their own standard variety of Dutch, but the way these standard varieties are positioned compared to non-standard varieties differs significantly (Geeraerts and Van de Velde 2013). In the Netherlands, colloquial language use tends to be linguistically closer to the standard variety. In Flanders, by contrast, a relatively delayed standardization history led to a situation of hyper standardization (Van Hoof and Jaspers 2012), in which the standard language enjoys considerable prestige but is seldom used apart from in the most formal situations. The colloquial varieties of Dutch in Flanders—while highly diverse—tend to be further removed from the standard variety than in the Netherlands. While both regions are characterized by considerable dialect levelling, this is particularly the case in the Netherlands.

A T/V distinction in second-person singular pronouns is found in Standard Dutch in both the Netherlands and Flanders: in subject form, u is the formal pronoun, while je is the informal one. Belgian Dutch also has the colloquial second-person pronoun ge which is found in both supraregional colloquial varieties of Belgian Dutch and certain local dialects (Vandekerckhove 2005). For the subject forms, both je and ge are unstressed forms and have a stressed equivalent, jij and gij, respectively. In Standard Dutch, this opposition between stressed and unstressed is also found for object (je and jou) and possessive (je and jouw) T forms. For colloquial Belgian Dutch, the object and possessive forms coincide with the respective V pronouns u and uw.

Table 1. Pronominal system for the second-person singular in Standard Dutch and colloquial Belgian Dutch.

	Standard Dutch			Colloquial Belgian Dutch
	T Pronoun		V Pronoun
	Unstressed	Stressed		Unstressed	Stressed
Subject	je	jij	u	ge	gij
Object	je	jou	u	u
Possessive	je	jouw	uw	uw

gives an overview of the second-person singular pronominal system in Standard Dutch and colloquial Belgian Dutch. The colloquial ge form is generally used in spoken informal language, as well as certain genres of highly informal writing like chatspeak (cf. Plevoets et al. 2008; Hilte et al. 2020), and hence shares indexical values with T forms. Despite occurring for stylistic purposes in marketing contexts (Van Gijsel et al. 2008), it is not usually found in the type of professional communication relevant to this paper, as this type of communication is normally written in Standard Dutch. Hence, we limit the discussion here to the T/V pronouns in the Standard Dutch system.

Research on T/V pronouns in Standard Dutch has predominantly focused on the Netherlands (Vismans 2015). Studies have documented the generally changing (reported) use of these pronouns throughout the twentieth and twenty-first century (e.g., Van den Toorn 1977; Van den Toorn and Vermaas 1988; Vermaas 2002; Vismans 2013a, 2013b, 2018). The most recent work confirms that the V form u is used in formal situations characterized by a larger distance between interlocutors, especially in educational and business contexts, whereas previously it was used to express an unbalanced power relationship between speaker and addressee (Jansen and Janssen 2005; Vismans 2013b; Vismans 2015; but see Levshina 2017 for recent data indicating social status differences as a dominant factor in determining the choice of u). The T form je is, as usual in T/V languages, reported to be used in exchanges where interlocutors have more common ground or are considered to be part of the same social group. The T pronoun is furthermore increasingly used in service encounters and is generally advancing compared to the use of the V pronoun (Vismans 2019; den Hartog et al. 2022; de Hoop et al. 2023). Research in Flanders is mostly limited to work on the competition between the Standard Dutch u/je system and the endogenous colloquial ge system in spoken Dutch (e.g., Deprez and Geerts 1977, 1980; Vandekerckhove 2004, 2005; Plevoets et al. 2008; Van De Mieroop et al. 2016; Zenner and Van De Mieroop 2021; Vismans 2015; De Dijn et al. n.d.). Recent work focusing on the use of the standard T/V system in Flanders is largely missing with the exceptions of Vismans (2007), Vismans (2013b), Oosterhof et al. (2017), and den Hartog et al. (2022), all of which take a comparative perspective between Belgian and Netherlandic Dutch (cf. infra). This bias in research focus on the variation between the standard and colloquial system rather than within the standard system for Dutch in Flanders reflects the firm presence of the topic in public debate.

If we turn our attention to what is known about Dutch T/V pronouns specifically in professional communication, work has been undertaken on a variety of genres, but again mostly with a focus on the Netherlands. Vismans (2019) reports on the negotiation of T/V forms in email correspondence between unacquainted writers. The study finds that the V form u is the default option at the start of email exchanges. Similarly, de Hoop et al. (2023) report evidence from an experimental study in which participants show more positive attitudes towards the use of u in emails relating to a job application procedure regardless of whether the email invited the addressee for an interview or rejected them, and regardless of participant profile (e.g., age, gender, level of education). This status of u as the unmarked default in professional interactions is echoed in a study by Jansen and Janssen (2005) that measures the appreciation of public information brochures using the V form u or the T form je. However, in line with the results reported in van Zalk and Jansen (2004), the authors do point out that the perception of u vs. je depends on interactions between the topic and the target audience of the text. Research on advertising banners in newspapers reports something similar to topic-based variation, namely that the choice of the V form u vs. T form je depends on the industry of the advertising company: the services industry tends to use u more often, while je is the pronoun of address of choice in entertainment and ICT advertisements (Vismans 2013a). Another study on T/V pronouns in advertising indicates a positive impact on the appreciation of an advert when the T form is used, but no effect of pronoun use on attitude towards the advertised product, purchase intention, or product price estimation was found (Schoenmakers et al. 2024). A final set of studies investigating T/V pronouns in Dutch language professional communications gear the attention towards a comparison between Belgian and Netherlandic Dutch. Generally, results are mixed. Vismans (2013b) reports no differences in the use of pronouns between the Netherlands and Flanders in his work on advertising banners. Similarly, den Hartog et al. (2022) who studied pronoun use in job adverts for multinational companies found that both Dutch adverts from the Netherlands and Belgium show a strong preference for T pronouns. Vismans (2007) by contrast does find regional variation in the use of T/V pronouns in job advertisements for highly educated job seekers, as well as variation dependent on industry. Overall, Flemish adverts use the V form u more, an effect that is driven by the financial–legal industry and the public sector (i.e., government, health, research, and education). Vismans frames these branches of industry as more conservative, especially in Belgium. The higher use of V forms in job advertisements in Belgium compared to the Netherlands is confirmed by Oosterhof and colleagues (Oosterhof et al. 2017).

To sum up, research on T/V variation in pronominal address in Dutch has mostly focused on the Netherlands and predominantly consists of work on (reported) usage. Studies including T/V variation in Dutch-speaking Belgium are scarce, just like research on the perception and evaluation of these pronouns. Findings relating to usage are mixed, but for Belgian Dutch the current state of the art seems to suggest that the V form u is the default option for written corporate communication, especially in more conservative sectors. Studies on the evaluation of the V form u vs. T form je in Flanders is lacking—to the best of our knowledge—but research in the Netherlands overall indicates that u is slightly more appreciated in professional copy (Jansen and Janssen 2005; de Hoop et al. 2023; but see Schoenmakers et al. 2024 for mixed results in an advertising context). Yet, more research on the perception of Dutch T/V variation could provide insights into whether this pronoun choice is salient to readers and hearers and in that sense should be of concern to professional copy writers.

2.3. Relative Salience of T/V Pronouns

The salience of a linguistic feature is a notoriously difficult concept to define (see, e.g., Zarcone et al. (2016) or Jaeger and Weatherholtz (2016) for reviews that aim to disentangle various interpretations). It is usually taken to refer to the fact that a feature is prominent in some way which may facilitate noticing (Kerswill and Williams 2002; Vandekerckhove and Ghyselen 2017; Jaeger and Weatherholtz 2016; but see for instance Rácz 2013 and Boswijk and Coler 2020 for more in-depth critical appraisals of the concept of salience in (socio)linguistics). This link with noticing is particularly relevant in sociolinguistics where it has been invoked as the crucial distinction between two types of sociolinguistic features: stereotypes and markers (Jaeger and Weatherholtz 2016). The former receive metalinguistic commentary and undergo style shifting, while the latter remain unnoticed by linguistic laypeople (Labov 1972). Pronominal address in Belgian Dutch has typically been classified as a stereotype in the Labovian sense (Van De Mieroop et al. 2016): language users are aware of the variation and there is metalinguistic debate on their use and social meaning in various contexts. As touched upon above, this prominence of pronominal address in public debate likely has its roots in the delayed standardization history of Dutch in Belgium, but deserves more support from empirical research.

While it is sometimes argued that linguistic features may be salient due to intrinsic qualities, often a feature will be prominent because it stands out in a certain context due to its unexpectedness (Vandekerckhove and Ghyselen 2017, p. 249; Rácz 2013; see Zarcone et al. 2016 for a discussion of the link between the concept of salience and that of surprisal). According to the latter view, salience is an inherently relative concept (Rácz 2013; Boswijk et al. 2020). Contextual elements that have an impact on how prominent a linguistic feature is could be social in nature: a feature may be salient if it is (un)expected based on certain speaker information like regional background (e.g., Niedzielski 1999; McGowan 2015). The unexpectedness may also be triggered by the linguistic context. Research on genre and register has shown extensively that certain types of discourse tend to exhibit particular bundles of co-occurring linguistic features (Biber 1988, 1995, 2019; Biber and Conrad 2005; Heylighen and Dewaele 2002; Koppen et al. 2019). Hence, if a feature is used that violates the expected combination of linguistic characteristics of a text, it may become salient in that text. This line of research has extensively documented linguistic characteristics co-occurring in formal and informal types of discourse. More formal genres and registers show a higher degree of informational density and typically contain linguistic features that serve an informational function, such as high frequencies of nouns, attributive adjectives, articles and prepositions, a high type/token ratio, and longer words (Biber and Conrad 2005; Heylighen and Dewaele 2002; Koppen et al. 2019). More informal genres and registers are more involved and are characterized by a higher degree of affection, empathy, and subjectivity (Heylighen and Dewaele 2002; Vis et al. 2012; Vliegen 2014). To that end, discourse of this type often exhibits high frequencies of (first- and second-person) pronouns, adverbs, verbs, interjections, and direct reported speech (Heylighen and Dewaele 2002; Vis et al. 2012; Biber and Conrad 2005).

To sum up, based on previous work, one could assume pronominal address in Belgian Dutch to be a sociolinguistic stereotype and hence be highly noticeable to language users. However, this assumption overlooks the fact that salience is not just an intrinsic quality of a linguistic feature, but also a contextual phenomenon: a feature’s salience depends on the situation it appears in and which other linguistic features are present in that context. In this paper, we prioritize such a contextualized approach to studying the salience of T/V variation in Standard Belgian Dutch. In particular, we aim to investigate to what extent participants with and without a professional background notice T/V changes in B2C communication and how this compares to other markers of informal discourse, viz. intensifying adverbs, exclamation marks, and English loanwords.

3. Research Questions

The first research question this study aims to address is as follows:

• RQ1: To what extent do Belgian Dutch language users notice pronominal variation in B2C communication?

As pointed out above, business communication is usually seen as a formal genre and studies on register, like Biber and Conrad (2005), have been able to put professional letters on the formal side of the continuum based on the linguistic features that characterize them. As a result, a reader may find informal features more unexpected in such texts and hence more salient, which increases their chance of being noticed. However, given the recent shift towards a more egalitarian work order, this preference for a formal variant may be called into question and we may even wonder whether pronoun choice is noticed at all. To better understand the impact of this evolution on the perception of pronouns of address in Belgian Dutch B2C communication, we further investigate the impact of additional informal features (intensifying adverbs, exclamation marks, and English loanwords) on the noticing of these pronouns. To that effect, we propose a second research question:

• RQ2: to what extent do other informal markers help notice a shift from V to T forms?

As discourse types are characterized by sets of co-occurring linguistic features, the informality of a text and hence the salience of an informal feature in that text can be hypothesized to depend on the features it co-occurs with in a text. This means that if we want to better understand how salient T forms are in business communication, we need to take into account their linguistic context. To that end, we aim to measure the impact of incrementally adding more informal features to the same text on the noticing of T/V pronouns. More specifically, we in turn add three features that the literature has characterized as informal. First, exclamations and the use of exclamation marks is usually advised against in more formal styles of writing (e.g., Chang and Swales 1999; Hyland and Jiang 2017; Lee et al. 2019). Second, intensifiers (for Dutch, e.g., erg, veel, or for English, very, strongly) have been associated with informal writing (e.g., Vis et al. 2012; Vliegen 2014). And finally, research has shown that the use of English words in Dutch is typical for informal styles as they have been shown to be used in situations with high involvement and emotion (Zenner et al. 2016) and are associated with teenagers’ online informal conversations and youth language more generally (Hilte et al. 2020). Including these additional features will allow us to explore the relative status of T forms in an increasingly informal text. It could be hypothesized that the presence of more informal features increases the chance of noticing other informal features. We could in turn expect that the more informal features a text contains, the more the text as a whole becomes informal and the unexpectedness and hence salience of the features decrease, leading to less noticing of T forms. Further, one could also put forward alternative scenarios in which for instance the most informal feature would stand out the most, perhaps directing attention away from T forms. These two hypotheses taken together make it more reasonable to assume that more informal features will lead to less noticing of T forms.

• RQ3: to what extent is noticing influenced by linguistic socialisation: do language users working in a language-oriented profession notice T/V shifts more readily than other language users?

RQ3 finally links up with the social factors that may play a role in salience, as pointed out above. It has been shown that language perceivers take into account social information about the speaker (e.g., Sumner et al. 2014; Niedzielski 1999; Hay et al. 2006). Perceivers can do that because they build on experience with socially meaningful language variation (McGowan 2016). This means that the background and socialisation of the perceiver is a factor that may influence noticing. In this study, we aim to investigate to what extent professional linguistic socialisation plays a role in the noticing of T/V pronouns in B2C communication. It could be the case that language users working in a language-oriented profession are more aware of the choice of pronouns of address as they have been taught more explicitly which is appropriate to a certain genre. They may also be more sensitive to language-related public debate, or they may be trained to pay more attention to linguistic choices in general. We have presented pronominal address as a linguistic feature that is above the level of awareness in Flemish society and regularly commented on, but it may not be equally noticeable for all readers/hearers.

Below, we present an experimental study designed to answer these research questions concerning the noticing of T/V pronouns in Belgian Dutch business communications in relation to variation in the linguistic context and the socialisation of the language perceiver.

4. Method

4.1. Design

To answer the research questions, we measured the perception of T/V pronouns in written corporate communications using a three-phase experiment followed by a sociodemographic questionnaire. In essence, the experiment offered participants a B2C information letter twice, with small manipulations between the versions. We then gauged whether participants noticed these manipulations.

• Phase 1: B2C information letter Version 1.
  • Phase 1a: Reading Version 1 under time pressure.
  • Phase 1b: Answering check questions.
• Phase 2: B2C information letter Version 2.
  • Phase 2a: Reading Version 2 under time pressure.
  • Phase 2b: Answering check questions.
• Phase 3: Salience test.
• Phase 4: Debrief and sociodemographic questionnaire.

In the initial phase, participants were presented with a first version of a text representing corporate B2C communication: an information letter from an insurance company outlining the advantages of the insurance packages chosen by the customer (cf. Section 4.2). Respondents were asked to read the text under time pressure, based on the average time a sample of pretest participants needed to superficially process the text. This way, all participants engaged in a fairly similar and nonchalant way with the text (phase 1a).

In phase 1b, participants were asked four easy questions to check whether they had actually read the text. Only participants who responded correctly to the questions could move on to the next phase of the experiment. Those respondents were told that they would see the same text again and were asked to read it again as fast as possible in order to answer some more questions, emphasizing the importance of reading the text again. The text they were shown in phase 2a was a manipulated version of the initial text containing a number of linguistic differences, including a shift in the use of T/V pronouns and, depending on the condition of the experiment (cf. below), up to three additional informal linguistic features.

After reading the manipulated text in phase 2a, participants went on to phase 2b in which they were asked to indicate whether they actually read the text a second time. Only those who indicated they did could continue to the third phase. In phase 3, it was explained to participants that the second text they read was actually not the same as the first text and they were asked whether they noticed any differences and if so, if they could describe those differences. At the end of the study, participants received information on the precise nature of the differences and were asked to provide basic sociodemographic information.

Participants were randomly assigned to one of five conditions. In each of those conditions, the number of linguistic differences between the text in phase 1a and the text in phase 2b increased with incrementally more informal features.

Table 2. Overview of the conditions in the between-subject experimental set-up.

Condition	Version 1 (Phase 1a)	Version 2 (Phase 2b)	N
C1	V pronouns	T pronouns	50
C2	T pronouns	V pronouns	53
C3	V pronouns	T pronouns + Intensifying adverbs	55
C4	V pronouns	T pronouns + Intensifying adverbs + Exclamation marks	49
C5	V pronouns	T pronouns + Intensifying adverbs + Exclamation marks + English loanwords	72

summarises the five conditions. In conditions 1 (C1) and 2 (C2), the only difference between the texts was the use of T or V pronouns. In condition 3 (C3), the T version of the text additionally contained intensifying adverbs. Condition 4 (C4) added exclamation marks to those features. Condition 5 (C5) finally contained the maximum of four informal features, adding English loanwords to the list. Note that to control for order effects, C1 and C2 were counterbalanced for whether the text containing V pronouns was shown first or whether the text with T pronouns was shown first. In C1, C3, C4, and C5, the text with V pronouns was used as the baseline text in phase 1a of the experiment given that one would traditionally expect V pronouns in formal types of communication such as a B2C information letter.

4.2. Materials

The texts shown in phases 1a and 2a of the experiment were based on an authentic letter from an insurance company. A number of adaptations were made to make the letter suitable for the experiment. First, the letter was anonymised to avoid effects of familiarity or existing associations with the insurance company. This meant that the name of the company and the agent, as well as the graphic identity of the company, were removed. Next, the letter was shortened from 437 to 189 words in order not to make the experiment too long and lose participants’ interest. Finally, the text was adapted to contain the linguistic features characteristic of each condition. The initial letter contained 14 V pronouns of address (7 of subject form u, 2 of object form u, and 5 of possessive uw; see Table 1 for an overview of the Belgian Dutch pronominal system) which were changed to T pronouns. The additional informal features were added to the letter to construct Version 2 for C3 to C5. Below, (1) illustrates the text manipulations, giving all possible manipulations (underlined) in the second paragraph of the body of the letter corresponding to Version 1 and Version 2 in C5, with the pronominal variants marked in bold. See Appendix A for the full text in all different conditions.

1.

Sleutels verloren? We zorgen voor een slotenmaker . |! Wenst | wens u|je een video-expertise? Dat kan, u|je filmt de schade met uw|je smartphone, de expert kan de opgelopen schade vanop afstand vaststellen en wij kunnen uw|jouw schade Ø |véél sneller vergoeden. In sommige gevallen is een herstelling in natura mogelijk. Wij zorgen er dan voor dat een vakman de schade komt herstellen, en we betalen de rekening rechtstreeks aan de hersteller. U|Jij betaalt niets. Dat bespaart u|je een hoop zorgen . | ! . Ook goed om te weten| By the way : we houden in onze tussenkomst rekening met de nieuwe bouwnormen.

Translation: ‘Keys lost? We arrange a locksmith . | ! Would you like video-based expert advice? No problem, youV | youT film the damage using yourV | yourT smartphone, the expert assesses the damage remotely and we will be able to reimburse youV | youT Ø |much faster. In certain cases we can offer you repairs in kind. We arrange a professional who will assess and repair the damage, and we pay the bill directly. YouV | YouT do not pay anything. That saves youV | youT a lot of worries. | ! By the wayDU | By the wayEN: we take into account the new building regulations in our intervention.’

In order to consolidate the hypothesized degree of informality of the additional informal linguistic features used in C3 to C5, a pretest was conducted (N = 6) in which participants rated which variant was the most informal for each variable (i.e., T vs. V pronoun, exclamation mark vs. full stop, intensifier vs. no intensifier, and English loan vs. Dutch alternative). This confirmed that the V pronoun was perceived as more formal than the T pronoun and that the use of intensifiers, exclamation marks, and English loans were perceived as more informal than the absence of these features. A follow-up task in which respondents ranked the informal variants in order of formality showed that English loans were seen as most informal, followed by exclamation marks and intensifiers, respectively. Further care was taken to select linguistic features that were easy to name for linguistic laypeople, so that if they noticed them in the salience test in phase 3, they would be able to verbalise what they had noticed.

4.3. Instruments

The instruments in this study consisted of two control questions in phases 1b and 2b, the salience test in phase 3, and the sociodemographic questionnaire at the end of the study in phase 4. The control questions in phases 1b and 2b aimed to detect participants who did not read the text at all, or with insufficient attention. As explained above, in phase 1b this was undertaken by directly asking participants if they had managed to read the entire text and by using content questions about the letter (see Appendix B). For phase 2b, this strategy was not an option to verify whether participants actually read the text a second time, given that the content of the text remained unchanged between phase 1a and 2a. Hence, we had to rely on participants’ honesty to directly report whether they completed the second reading. Of course, this strategy is prone to social desirability bias, but we believe that bias to be somewhat attenuated by the fact that the study was anonymous and not taken in the presence of the researchers.

In phase 3, participants were first told that the texts they read in phases 1a and 2b were not exactly the same. They were then asked in an open-answer field to describe any differences they noticed as precisely as they could. Participants who did not notice any difference could also indicate this and move on to the final part of the study (phase 4).

The sociodemographic questionnaire contained questions about participants’ gender, age, residence, native language, level of education, and whether participants were in a language-related occupation (e.g., teacher, editor, journalist, translator). The latter question served to measure professional linguistic socialization, and hence we prioritized the participants’ own assessments, asking them whether they were engaged with language in a professional capacity.

4.4. Procedure

The experiment was conducted through the survey platform Qualtrics. Data collection was anonymised, and this was clearly communicated to participants at the start of the experiment. The study was presented to participants as research on reading skills. It was not possible to give full disclosure on the exact nature of the research questions at the start of the study, as this would have interfered with its aim. In phase 1a, participants had 52 s to read the information letter in Version 1. The maximum display length was determined after pretesting to make sure that the vast majority of readers applying the targeted casual reading style would have enough time to read the entire text. Participants who took more than the allotted time were screened out and debriefed about the reason they were screened out. Those who read fast enough went through to phase 1b. Participants who did not pass the control questions in this phase were screened out, again with the necessary explanations about why the experiment ended there for them. Those who read carefully enough went on to phase 2a, where they read Version 2 of the text. To encourage participants to read the text again, the button to move on to the next page phase only appeared after 52 s. In phase 2b, participants were asked to honestly indicate whether they did indeed reread the text. Those who said they did not were screened out with the necessary debriefing. Those who said they did advance to the salience test in phase 3 and afterwards to the sociodemographic questionnaire in phase 4. At the end of phase 3, after completing the salience test, participants were debriefed about the actual goal of the study.

4.5. Participants

Participants were recruited in the first half of 2019 through the personal networks of the authors on social media, as well as through the professional network of the fourth author. Participation was voluntary. Originally, 543 respondents started the experiment. Of those, 366 made it to the end of the study, 84 dropped out early because they did not read one of the texts fast enough (cf. Section 4.4 Procedure on the reading time threshold), and 93 were led to the end of the survey because they failed to respond correctly to the reading check questions in phase 1b.

Out of the 366 complete participations, data from 87 additional respondents were disregarded for one of the following reasons. We excluded data from participants whose native language was not Dutch (N = 5) and from participants who were not originally from Flanders or Brussels (N = 3). As we wanted to ensure a sample that was as homogeneous as possible, we excluded two groups who were not numerous enough to be well represented in each of the conditions: participants who were not highly educated (N = 45) and participants who were not part of what is traditionally considered the active population (i.e., over 65 years old, N = 4). Finally, participants from the western-most province of Flanders, West Flanders, were not included in the analyses (N = 30). The reason for this last decision is that the endogenous T form of the second-person pronoun singular is different in the West-Flemish dialect area compared to the rest of Flanders: the T form used in several (mainly coastal) dialects in this area is the same as the T form in the standard language, namely jij, while in the rest of Flanders gij forms are more common (see Devos and Vandeweghe 2003). It is not inconceivable that having a different endogenous T form could influence the perception of its social meaning in terms of formality in West Flanders compared to the rest of Flanders. As a result, the salience of the use of T/V pronouns in formal communication may differ for participants from this region, but verifying this hypothesis is beyond the scope of this study.

Given these exclusion criteria, the sample used for analysis consisted of N = 279 participants. Participants belonging to various demographic groups were distributed evenly over the conditions: chi square tests showed no significant skew in the distribution of participants from different regions (χ² = 10.46, df = 16, p > .1) or gender (χ² = 3.82, df = 4, p > .1). A one-way ANOVA did not indicate differences in mean age over the conditions (F = 1.28, df = 4, p > .1). There was furthermore no significant difference in the distribution of respondents who used language professionally (e.g., teachers, editors, journalists, translators) and those who did not over the conditions of the study (χ² = 3.19, df = 4, p > .1). For a detailed breakdown of participant demographics per condition, see Appendix C.

4.6. Analyses

The data provided by participants in the salience test in phase 3 were coded independently by two researchers (one of which was the first author) to indicate whether participants mentioned the presence of each of the linguistic features under study (i.e., T/V pronouns, intensifiers, exclamations marks, and English loanwords). For instance, the response in (2) was coded for noticing intensifiers, exclamation marks and English insertions, but not T/V pronouns.

2.	Er kwamen enkele Engelstalige termen voor, zoals “by the way” en “oldschool”. Daarnaast werden er ook accenten geplaatst op bepaalde woorden, zoals “héél” en “véél”. Ik denk ook dat er meer uitroeptekens werden gebruikt in de tweede tekst [id024, C5]
	Translation: ‘There were a few English terms, like “by the way” and “oldschool”. Additionally accents were placed on certain words, like “héél” and “véél”. I also think that more exclamation marks were used in the second text.’

We only coded for features that actually appeared in the condition in question. In example (3), the participant correctly notices the use of English words, but also reports that a procedure for repairs was explained differently which was not the case. In this example, the imagined difference was not coded for.

3.	Er wordt nu over oldschool gesproken. De procedure voor de herstelling door een vakman leek ook duidelijker uitgelegd. [id051, C5]
	Translation: ‘Now oldschool is mentioned. The procedure for the repairs by an expert seemed to be explained more clearly.’

The coding strategy was conservative as to avoid an incorrect interpretation of participant responses. For example, if a participant mentioned that they noticed that the second version of the text contained ‘modern words’, as in (4), this most likely referred to the English loanwords in C5, yet this was not coded as such to avoid overinterpretation. Ambiguous responses like these were, however, rare, and in most cases it was clear whether a participant identified the difference(s) correctly or not.

4.	Modernere woordenschat [id004, C5]
	Translation: ‘More modern vocabulary’

Analyses were carried out in R (v4.3.0, R Core Team 2023) using stats (R Core Team 2023) and emmeans (Lenth 2023) as the main packages.

5. Results

To tackle RQ1, we specifically verify whether participants reported noticing a change in T/V pronouns between the two texts they were offered. Overall, we see that, across conditions, 17 of our 279 participants (6%) report noticing such difference (see

Table 3. Overview of participant numbers and proportions reporting noticing a difference in T/V pronouns between the two text versions.

	Language Professional			Other Professional Backgrounds			Total
Condition	N T/V Noticed	N T/V Unnoticed	% Noticed	N T/V Noticed	N T/V Unnoticed	% Noticed	N T/V Noticed	N T/V Unnoticed	% Noticed
C1	3	24	11%	2	21	9%	5	45	10%
C2	2	18	10%	1	32	3%	3	50	6%
C3	0	25	0%	2	28	7%	2	53	4%
C4	2	20	9%	0	27	0%	2	47	4%
C5	4	32	11%	1	35	3%	5	67	7%
Total	11	119	8%	6	143	4%	17	262	6%

). To answer RQ2 and RQ3, i.e., to determine to what extent noticing was impacted by the presence of additional informal language features and participants’ linguistic socialization, responses were modelled using a generalized linear model. The model included the reporting of noticing a change in T/V pronouns (yes vs. no) as the binary response variable, and condition (C1 vs. C2 vs. C3 vs. C4 vs. C5), language socialization of the participant (language professional vs. other professional backgrounds), and their interaction as fixed effects. No significant effects of either condition, language socialization, or their interaction were found (see Appendix D for details on the model). Given the skew in the distribution of the response variable, we additionally verified the impact of condition and professional background on T/V change noticing using conditional inference trees (Tagliamonte and Baayen 2012). A lack of branching in the tree confirmed the conclusion that neither factor seemed to impact on the degree of noticing a pronoun change between the text versions.

Although strictly speaking not within the scope of our research questions, we supplemented the above analysis with a further analysis of the noticing of the other linguistic features to put the noticing of T/V pronouns in perspective.

Table 4. Overview of participant numbers and proportions reporting having noticed a difference in one or more of the four features under study and those who did not.

	Change Noticed		Change Unnoticed
	N	%	N	%
T/V pronouns	17	6%	262	94%
Intensifiers	18	10%	158	90%
Exclamation marks	6	5%	115	95%
English lexical items	45	63%	27	37%

gives an overview of the proportion of participants noticing each of the four linguistic features under study. Overall, just like T/V pronouns, intensifiers and exclamation marks are rarely reported as differences between the two texts: on average, 10% of participants report the intensifiers, and 5% the exclamation marks. These figures are in the order of the 6% noticing of the T/V pronouns. By contrast, the English lexical items are noticed by 63% of the participants in C5, the only condition including this feature. A chi square test confirms a significant association between noticing and the type of features for Table 4 (χ² = 172.77, df = 3, p < .0001, Cramer’s V = 0.52). Interestingly, this considerably higher degree of noticing of English lexical items in C5 did not increase or decrease the degree of noticing of the T/V pronouns in this condition (cf. lack of effect of condition in the model reported in Appendix D). In a series of generalized linear models for each of the additional features separately,2 we also found no significant differences in noticing for any of the features for language users working in a language-oriented profession compared to other participants.

6. Discussion and Conclusions

The experiment suggests that changes in T/V forms in pronominal address in Dutch professional communication are not very salient: less than 1 in 15 respondents noticed that the pronouns of address varied between two versions of the same information letter they read in the study. Moreover, there was no difference in noticing between participants in a linguistic profession and those without a language-oriented professional background. Interestingly, boosting the informality of the text by adding other informal linguistic features did not impact the salience of T/V variation. Two of these additional features, intensifiers and exclamation marks, went virtually unnoticed as well. By contrast, English lexical items, the most informal feature that was added in C5 alongside the T pronouns, intensifiers, and exclamation marks were picked up significantly more frequently, with over 60% of participants noticing their appearance in the second version of the information letter. These results lead us to formulate negative answers to RQ1, RQ2, and RQ3.

These results raise two questions that require further discussion: (1) Why do only few participants notice the difference in pronominal address? (2) Why do English lexical items get noticed? Starting with the first question, we can consider several potential explanations. A first hypothesis draws on research in the field of visual perception: we may be confronted with a case of change blindness owing to the design of the experiment (Simons and Levin 1997). This phenomenon is defined in psychology as “the surprising failure to notice seemingly obvious changes from one moment to the next, […] people are ‘blind’ to the changes between what was in view moments before and what is in view now,” (Chabris and Simons 2010, pp. 54–55). While work on change blindness does not tend to focus on the type of linguistic changes relevant for this study, and hence further research would be required to give a decisive answer regarding its relevance here, the phenomenon does offer an interesting avenue to reflect on why participants did not notice the change in T/V pronouns in the experiment designed here. Many theories accounting for change blindness suggest that it occurs for elements that are not significant for the interpretation of a visual scene (Kentridge 2015). Likewise, the linguistic changes in our experiment do not change the understanding of the text as a whole. Another feature of our experimental design that may facilitate change blindness is the amount of time between seeing the first and the second text. Research suggests that the longer the time between two scenes, the less likely a visual change will be noticed (Hollingworth 2004). The comprehension check (phase 1b) in between our two text versions may hence have contributed to change blindness. Yet, if we invoke change blindness as an explanation for (1), the question remains (2) why the English items did get noticed. Future research could explore when and to what extent textual change blindness occurs, e.g., by including other textual changes than those implemented here, by varying the attention drawn to possible changes between text versions in the instruction phase, varying the amount of time included between variants in the reading itself, or in the elicitation of the perceived differences in the test phase.

Second, de Hoop et al. (2023) and Schoenmakers et al. (2024) suggest that the salience of T pronouns may be influenced by whether their stressed or unstressed forms are used. This certainly is a plausible suggestion, but we cannot confirm or reject this hypothesis based on the current study. We included a mix of stressed and unstressed forms which we felt were natural in the context of our information letter, and as such hope to have improved the ecological validity of the experiment. Future research could explore to what extent change between stressed forms may be picked up more easily than changes to unstressed forms. For instance, eye-tracking methods could be used to investigate which forms draw attention (compare, e.g., Hilberink-Schulpen et al. 2016, for the use of foreign languages in job advertising). Eye-tracking methods could also be used in between-subject set-ups where participants read only one version of the text to avoid reduced ecological validity by presenting the same (slightly altered) text twice.

A third alternative explanation for (1) may be found in Jansen and Janssen (2005), who suggest that the topic of the text in interaction with the attitudes of its readership towards that topic may impact the perception of T/V pronouns. Their study on text appreciation in relation to the use of T/V pronouns indicated that participants who are indifferent about the topic of the text are insensitive to the use of different pronouns in the text. It is not unlikely that the topic of the informational text on insurances in our study was not of particular interest to many participants. This indifference to the topic perhaps also led to less attention to pronouns of address. Additionally, the experimental setup reduced the ecological validity of the reading task, which may have also led to lower levels of engagement with topic and text, even for participants who are supposed to engage with text and language on a professional level. As such, the response to (1) may not (solely) lie in aspects of the experimental design, but may also be answered by looking into the respondent profiles of those who (do not) notice the T/V pronouns. While we were able to show that a professional linguistic background is not what sets these participants apart, we do not have any other data at hand or a sample large enough to investigate further individual differences in linguistic experience or cognitive and sociodemographic characteristics that may be involved in noticing T/V pronoun use. Future research could for instance look at the impact of (individual) reading styles and reading skills, involvement of readers with the text, or consider aspects of socio-economic background other than language orientation in participants’ professions.

The most likely explanation for (1), however, may be found in our initial appraisal of the formality level of either the text genre or the T pronouns. It could be that the professional information letter in B2C communication has recently come to be perceived as more informal than previously reported in the literature. As rightly pointed out by an anonymous reviewer, the fact that we use we/wij forms in the letter may have added to this more informal style. Hence, T pronouns would not be out of place in a genre that is not seen as overly formal. However, if the genre was perceived as genuinely informal, then we would expect a higher noticing of V pronouns, especially in C2, which was not the case.

More plausible is that the relationship between T and V pronouns in the Belgian context is not quite the same as the one in the Netherlands, which has been described more extensively in the literature and which led us to formulating our hypotheses. As discussed in Section 2, the Belgian linguistic landscape has an additional option for pronominal address, namely the colloquial gij system, which has been convincingly shown to be a stereotype in the Labovian sense (Van De Mieroop et al. 2016). Speakers are highly aware of its non-standard status, yet at the same time it is the most natural option for many, as for few Belgian Dutch speakers the standard T/V distinction is part of their native variety (Deprez and Geerts 1977; Vandekerckhove 2005). Hence, it may be the case for these speakers that the difference between the gij and jij system is much more salient than that between the jij and u system, with the latter two options being both perceived as standard and more formal, their difference in formality fractional compared to the difference with gij. This hypothesis that puts the social meaning of jij close to that of u compared to gij would be in line with recent anecdotal reports of changes in the pronominal system where je is becoming the new u, effectively taking over its function as a V form, leading to a T/V system with jij as the V form and gij as the T form. Indeed, the Belgian Dutch context is quite unique in offering speakers two T forms alongside a V form. If jij as a standard T form is undergoing change to take up part of the social work traditionally performed by V forms (see De Dijn et al. n.d.), and to as such set it apart from the colloquial T form gij, this may well explain why it did not stand out to participants in a formal information letter. A factor that could contribute to this levelling of the perceived formality difference between the Standard Dutch T and V pronouns is that the forms u and uw also feature as the oblique form in the gij system, meaning that the same linguistic forms are both used in informal and formal contexts. As can be seen in Table 1, the oblique form of the gij system is u, which formally coincides with the u forms in the Standard Dutch V system. This means that language users can produce formally identical variants in both formal and informal discourse, which could potentially attenuate the formality associated with the V system, reducing the difference with the T system (cf. De Dijn et al. n.d. on ambiguity of u forms in professional communication).

Let us now turn to the second open question: why were English lexical items noticed more readily than the other linguistic features? While this question goes beyond the scope of the original research questions of this study, it is a noteworthy finding that deserves further consideration. To explain the salience of the English items, one could argue from a visual prominence perspective that they are longer and perhaps in that sense visually more readily perceptible than exclamation marks or pronouns. This explanation, however, cannot account for the fact that the intensifiers were hardly noticed. The latter are also full lexical items which were furthermore emphasized visually by using stress marks. Another straightforward explanation concerns the shift in language: the English items might be more salient simply because they are in English (see also Hornikx and van Meurs 2020).3 Another explanation is that these items cross a certain threshold of informality so that even if the informational B2C letter has become less formal as a genre, they still stand out. The fact that the English loans rather than the other features got noticed by a significantly larger portion of the participants is in line with the pretesting of the informality of the linguistic features that informed the experiment. Note that a qualitative analysis of the responses offered in phase 3 of the core experiment suggests that many respondents who did not explicitly name the English loans (and hence are not included in the proportion of participants who noticed the change in the text) did report a change in the style of the text. Participants, for instance, mentioned that the second version of the text seemed more ‘informal’, ‘jovial’, ‘smooth’, ‘modern’, and ‘youthful’. Of course, we cannot claim with any certainty that these qualifications of the second version of the text are to be associated with the English loans, as all informal features or their accumulation in C5 potentially made the text more ‘informal’, ‘jovial’, and ‘smooth’. Yet, descriptions like ‘modern’ and ‘youthful’, which were absent in the other conditions, echo indexical values that have been associated with the use of English in Belgian Dutch. For instance, the perception study by Schuring et al. (2023) shows how the use of English lexical items in Dutch is linked more with (pre)adolescents than with older participants, and more with modern roles (e.g., rapper) than with traditional roles (e.g., ‘butcher’) or public roles (e.g., ‘mayor’). Likewise, in their summary of the associations linked to the use of English (in advertising), Hornikx and van Meurs (2020, pp. 99–100) identify ‘modernity’ (including values such as youth and future), ‘success’, and ‘internationalness’ as main dimensions.

As this research was inspired by questions arising from the field of professional communication, it seems only fitting to conclude this paper with an attempt at formulating more practical advice that can provide guidance to communication professionals. At first glance, one may be inclined to conclude that it does not matter a great deal whether one uses the T or V form in Belgian Dutch business communications, as most people do not notice the difference. The results show that shifts from T to V forms, or vice versa, are not noticed by the majority of respondents. However, caution is warranted and there may be more at play. First, it is clear that the ecological validity of an experiment like the one reported here is limited. This means that the results may not transfer readily to real life communicative situations. Second, linguistic research has not been able to determine what the role of noticing and salience is in the activation of indexical meaning. As pointed out in the introduction, the reason why companies and communication specialists are in doubt whether to use T or V forms boils down to the question of what indexical meanings these forms introduce in communication, which feeds into a company’s tone of voice which in turn is crucial in determining its public perception. Preston (2011), for instance, argues that noticing, be it conscious or subconscious, is a necessary condition for the activation of social meaning (i.e., the Contrastive Mandate). Drager and Kirtley (2016), by contrast, claim that social meaning activation is not dependent on any form of noticing. It may be clear that both linguistic theorising as well as empirical work on this topic are urgently needed. The same goes for further research on pronouns of address in Belgian Dutch, not just focussing on their perception as we have here, but in parallel with studies on evaluation and the production of T/V variation. For now, the best advice for companies seems to be to choose the pronoun of address with the social meaning that best fits the corporate tone of voice, but consider that many customers may not pick up on the difference.