‘It’ Is Not for Everyone: Variation in Speakers’ Evaluation of Sociopragmatic Pronouns in Limburgian

Abstract

This paper explores the different ways in which speakers of Limburgian think and feel about sociopragmatic pronouns in their dialect, in which women can traditionally be referred to with both ziej ‘she’ and het ‘she’ (lit. ‘it’). Previous research revealed variation between speakers regarding the use of het, which appears to be associated with differences in interpretation and evaluation. This study investigates this further by analyzing how individual speakers evaluate non-feminine pronouns for women. Our data show that many speakers have a relatively high level of awareness, discussing four key themes: (i) how female reference in Limburgian differs from that of Dutch; (ii) the appropriateness of using the pronouns for certain referents and/or in certain social situations; (iii) the various connotations the pronouns may have; and (iv) how they navigate between-speaker differences regarding pronoun evaluation in daily life. Importantly, this high level of awareness is not present in all speakers. These results indicate that the use or non-use of the pronoun het for female referents may be conditioned by cognitive automaticity for some speakers, and by a conscious, emotionally charged consideration about its desirability for others. Our findings, therefore, suggest the importance of integrating cognition and sociality in linguistic theorizing.

1. Introduction

While for speakers of English and many other languages it may seem obvious that you refer to a man with a ‘masculine’ pronoun (‘he’) and to a woman with a ‘feminine’ pronoun (‘she’), many languages use other reference systems. This includes languages in which there are no separate pronouns for different genders, and also systems in which reference to female referents contains sub-divisions not found for reference to male referents. The current article is about one such language, but it is important to note that such dynamics exist in many languages of the world, and therefore reflect a conceptual distinction people have apparently often felt is important to make.

When languages use more than one pronoun to refer to women, the different pronouns tend to encode social or pragmatic information about the referent, in addition to basic grammatical (‘3rd person’) and semantic (‘female’) meanings. Such information often includes an assessment of the relationship between the speaker and the referent (Aikhenvald 2016; Corbett 1991)—much like the politeness distinction many languages encode in contrasting two or more second person pronouns (Helmbrecht 2013)—and is referred to as ‘sociopragmatic gender’ (Busley and Nübling 2021; Steriopolo 2021). ‘Sociopragmatic gender’ refers to the use of a linguistic item of a specific gender to express social information which goes beyond the mere indexing of the referents’ gender or sex. The choice for certain gender forms can reflect specific social information about the referent or about the relationship between the speaker and the referent. The logical consequence, of course, is variation: in languages with socio-pragmatic gender, there are at least two pronouns that are both used for reference to women.

Limburgian (West-Germanic) traditionally has two options for pronominal female reference, in three grammatical roles: subjects (grammatically feminine ziej ‘she’, or grammatically neuter het ‘she’, lit. ‘it’.)1; direct and indirect objects (feminine eur ‘her’ or masculine hem/um ‘him’, which is also used for male referents); and possessives (feminine eur ‘her’ or masculine/neuter zien ‘her’, lit. ‘his/its’). In addition to personal pronouns, pronominal reference to people can also be realized with demonstrative pronouns; when the referent is a woman, this pronoun can be feminine die or neuter det (both forms literally mean ‘that’) for both grammatical subjects and objects (see Piepers et al. 2021 for a more elaborate sketch of the Limburgian pronominal system). Importantly, Limburgian is a minority language, and the feature of sociopragmatic gender is not shared with the contact language, standard Dutch (cf. Section 2.1 below). Sociopragmatic gender is, however, present in various other West-Germanic language varieties related to Limburgian (see, e.g., Busley and Nübling 2021; Martin 2019; Nübling 2015; Nübling et al. 2013). Some example sentences of the variation between two pronoun genders, taken from the DynaSAND corpus (Dynamic Syntactic Atlas of Dutch Dialects; Barbiers et al. 2006), are included below (note that the DynaSAND corpus does not provide an example of the masculine object form hem with a female referent):

• ziej/het haet zich pien gedaon
  ‘She(F/N) hurt herself’.
• Marie, die/det zoow zoget noots doon
  ‘Mary, she(F/N) would never do something like that’.
• Marie eure/ziene auto is kepot
  ‘Mary’s(F/M/N) car is broken’.

As the label ‘sociopragmatic’ indicates, the feminine and non-feminine forms are not interchangeable outright. In earlier research, we saw identifiable patterns in how they are used: both het and ziej are unmarked options for reference to girls and young women, but het is generally dispreferred for reference to an older woman—in such cases it was produced less frequently and judged as less acceptable. Older women were usually referred to with the feminine form ziej (Piepers et al. 2021). The choice of pronoun thus appears to be governed by a sociopragmatic constraint, similar to what has been described for various German dialects, Swiss German, and Luxembourgish (Nübling 2015). The same ‘age-effect’ surfaced in another study we did (Piepers et al. 2023): a young girl was more often referred to with het than teenage girls, who were in turn more often referred to with het than adult women (cf. Braun and Haig 2010).

Importantly, these studies also revealed that there were considerable differences between Limburgian speakers in their use of the pronouns: the socio-pragmatic constraint does not predict pronoun use completely. In the first study, 31.71% of the participants completely refrained from using het as a subject pronoun. Their disuse of het could not be linked to the speakers’ age, gender, regional origin, or educational background, suggesting other factors are at play. Among the remaining speakers, the use of non-feminine pronouns showed considerable variation, ranging from less than 10% to 90% of the time (Doreleijers et al. 2021). The second study, in which participants first described images and then translated Dutch sentences into Limburgian, again showed large amounts of variation: in the description task, 32.85% of the participants never used het; in the translation task, this was 66.18% (note that Dutch does not use het, so the translation primes never contained it). For both tasks, the other speakers ranged in their use of het from 1% to 100% (Piepers et al. 2023).

When the choice of pronoun is dependent on the speaker’s assessment of interpersonal relationships, it is to be expected that the choice will not always be crystal clear. Sociopragmatic gender assignment, therefore, can be expected to give rise to variation in everyday language use. This variation may simply originate from the vagueness of the underlying categorization, some people settling on form X and others on form Y for the same referent without necessarily conveying any social meaning. However, the different choices may also be indexical of different ways of judging appropriateness (see Eckert 2019, pp. 93–96, for a review of relevant studies in which the choice of linguistic element indexes the social group to which one belongs). One can imagine that a pronominal choice that encodes the speaker’s estimate of both the social distance and the distance in age to the referent and that in addition embodies a fundamental difference between talking about women and talking about men, holds the potential for developing ideological differences. The choice between pronouns might become imbued with socio-political meaning, and the fact that speakers differ in whether or not they use the non-feminine pronoun suggests that this may be a possibility in the modern Limburgian speech community.

Against this background, sociopragmatic gender and the choice of pronoun present an interesting research topic. Variation may be conditioned by sociolinguistic factors, such as regional origin, gender, or age, but also more locally meaningful dimensions. This paper explores the individual differences in evaluation and interpretation that may lie behind the variation between speakers in their use of third person pronouns for female referents in Limburgian. Building on the results of the previous research referenced above, which showed that the use of Limburgian pronouns depends on sociopragmatic factors but is likely also sensitive to opinions and value judgments, this paper zeroes in on how speakers understand and value this aspect of their language. Specifically, we investigate speakers’ individual evaluations of the appropriateness and connotations of non-feminine pronouns when referring to women. We explore how speakers view this feature both in pragmatic and socio-cultural terms and whether pronoun choice has become a sociolinguistic marker.

2. Background

2.1. Sociopragmatic Gender in Limburgian-Dutch Language Contact

Limburgian is a “nonunique, adjoining, cohesive indigenous minority language” (Louden 2020, pp. 818–19), which means that it is spoken in a concentrated geographical area spread across multiple countries. In all of these, it is a minority language; in this case, it is spoken in the Netherlands, Belgium, and Germany. Limburgian has no official status in the latter two countries, but in the Netherlands, it has been recognized as a regional language since 1997. It shares vocabulary and grammatical features with both Dutch and German, and can therefore be described as ‘straddling the border’ between Low and Central Franconian (cf. Bennis and Hermans 2013).

The research we report on focuses on Limburgian as spoken in the Netherlands. It is important to note that despite its status as an official regional language, there is no such thing as ‘standard Limburgian’: it has no prescriptive rules, and exhibits abundant regional and local variation. Limburgian is not used in educational settings; speakers learn Limburgian at home or in social settings, but not in school.

The Dutch province of Limburg is usually divided into six main dialect areas, all of which are mostly mutually intelligible (e.g., Cornips 2012). While all dialects within Limburg are ‘Limburgian’ in the geographical and political sense of the word, they are not all ‘Limburgian’ in the linguistic sense. The dialects in the northern parts of Limburg are classified as Kleverlandish (Low Franconian) and the southeasternmost ones as Ripuarian (Central Franconian, to be precise West Middle German). While Ripuarian shares with Limburgian the feature of sociopragmatic conditioning of third person pronoun choice for female referents, and specifically the use of a neuter subject pronoun (Nübling 2015), Kleverlandish does not. For this reason, ‘Limburgian’, in this paper, excludes Kleverlandish, and refers to those dialects that are spoken in the remaining five dialect areas. Notably, however, a lot of regional variation exists even within dialect areas: any ‘Limburgian’ speech or text is always an instance of a certain local dialect (Bakker 1997), and throughout their everyday life, speakers are exposed to a variety of different dialects and dialect combinations (Ramachers 2018).

While dialects in the Netherlands generally seem to be losing ground to the standard language (Versloot 2021), Limburgian is relatively stable: although older people reportedly are fluent more often than younger people, recent data do show that Limburgian is used by all social classes, both in domestic and public contexts (Schmeets and Cornips 2022). Still, Limburgian is under pressure from the national standard language Dutch, a language used daily by all speakers from early childhood on. In this sense, the sociolinguistic situation in Limburg mirrors that of many other European countries, where dialects are changing rapidly due to the omnipresence of the standard language (Auer et al. 2005).

Studies of bilingualism have amassed abundant evidence that bilinguals do not ‘switch off’ one language when using the other (e.g., Backus et al. 2011; Barking et al. 2022; Blumenfeld and Marian 2007; Travis et al. 2017): in fact, cross-language activation is “a prime example of language contact in the bilingual mind” (Kootstra and Muysken 2019, p. 1), and cross-linguistic influence is common. Though to the best of our knowledge there have been no systematic studies of Dutch influence on Limburgian in general (but see Giesbers 1986; Vousten 1995 for interesting case studies), it stands to reason that in the speech of almost all speakers, there will be many instances of word choice, word use, and grammatical choices that betray the impact of Dutch. If pronoun use is changing, this too may partly be due to Dutch influence. There is no sociopragmatic gender in Dutch: women are always referred to with the feminine personal pronoun (Audring 2009), and the neuter personal pronoun het reliably refers to inanimate referents, and hardly ever to human beings.

There is very little to no information available on Limburgian pronoun choice in the past. That is, we cannot know for sure whether avoiding non-feminine forms is a recent innovation or whether it was always possible (i.e., whether there have always been speakers who never used het; cf. Doreleijers and Swanenberg, forthcoming). It is also unclear whether the current rate of use of het, when socio-pragmatically appropriate, has lowered from 100% in the past to something lower nowadays for some speakers. Kats (1939, p. 118) wrote that “the neuter(!) form het, with the oblique case hem, is only to be used for a woman or a girl with whom both speaker and addressee are on a confidential and at least equal footing, but must be used in this case” (emphasis original; translation is ours). This suggests that there is a change in progress; i.e., that the optionality of the pronoun may have arisen recently, with increasing bilingualism (Swanenberg 2020).

Given the intense bilingualism, the stage is set for a continuous influence of the standard language on the regional language, meaning that speakers might well experience varying degrees of interference from their Dutch pronoun use. The Dutch denotational meanings of the pronouns (i.e., inanimacy in case of het ‘it’ and dat ‘that’, and masculinity in case of hem ‘him’ and zijn ‘his’) might affect how individual speakers interpret their Limburgian counterparts, invoking negative or awkward connotations when used to refer to women or girls. For example, already in the early 90s, Bakker (1992, p. 15) wrote that “inquiries with speakers show that some think het and det sound negative”, and that “a semi-native speaker of a central Limburgian dialect said to never use the informal [non-feminine] forms, because he thinks they sound contemptuous”. Bakker also noted that “some young people” use feminine forms, whereas “older and other speakers always” use non-feminine forms (Bakker 1992, p. 14). Follow-up research about thirty years later indicated that this division between ‘users’ and ‘non-users’ still exists, and, importantly, that there are now both younger and older speakers (i.e., the ‘young people’ from 1992) who do not use het (Piepers et al. 2021). It is possible that in the intervening three decades, this feature may have become more salient and the subject of metalinguistic discussion for some speakers. Negative connotations may cause some Limburgian speakers to avoid the non-feminine pronouns; eventually, this could spread across enough speakers to raise the impression that the language may be losing the use of these pronouns (cf. Aikhenvald 2003; Thomason 2001 for general accounts of cultural causes of contact-induced change; for specific accounts of contact-induced change in gender systems, see e.g., Clamons 1995; Gumperz and Wilson 1971). In the next subsection, we discuss the sociolinguistic implications of connotational changes.

2.2. Impact of Awareness and Evaluation on Selection of Linguistic Items

Given that the variation between Limburgian third-person pronouns for female referents is governed by sociopragmatic considerations, it is likely that personal preferences and opinions—i.e., attitudes—may at least partly be responsible for the choices people make. The relevant question becomes one of metalinguistic awareness, i.e., “the ability to reflect upon and manipulate the structural features of spoken language, treating language itself as an object of thought, as opposed to simply using the language system to comprehend and produce sentences” (Tunmer and Herriman 1984, p. 12). That is, what do speakers of Limburgian know, think, and feel about their language—more specifically, about sociopragmatic pronouns—and how does this shape their use of these pronouns (cf. Giles and Coupland 1991; Ladegaard 2002)?

When attitudes are widely shared, variation is likely to reflect classical sociolinguistic background factors (age, ethnicity, gender, etc.), but since these factors are not always of prime local relevance in social stratification, this will not always be the case. This is because “language ideologies are grounded in social experience, which is never uniformly distributed throughout polities of any scale” (Kroskrity 2004, p. 503). Resulting from differences in social and linguistic experience, the same word may not have the same meaning to different speakers (cf. Ramsey 2022), and speakers do not necessarily think about what is ‘good’ language in the same way—rather, their “perceptions diverge, in socially important ways” (Bijvoet and Fraurud 2015, p. 20). In such situations, newer and older ideologies might simultaneously be present, and their contrast may be visible in the linguistic choices people make and in the things they may have to say about it. One sign that a speech community is going through this early phase is the absence of much or any overt ideological discussion about the feature in question, in public or in private communication: if the feature is not particularly salient in the public consciousness, it cannot give rise to an ideological stance about it. In the current paper, we investigate whether the feature we focus on is starting to gain in salience and is giving rise to the development of an ideology.2

Against this theoretical background, the Limburgian case is an interesting one. On the one hand, previous research showed that the pragmatic constraints are fairly well entrenched across the speech community and that the pronominal choice often seems to carry no additional social meaning, in the sense of marking ideological stance or group membership. However, since there is considerable variation in the use of non-feminine het, including competition from the feminine pronoun where conventionally the non-feminine pronoun would be acceptable and even complete non-use, it might be, at least for some speakers, a marker in the Labovian sense, i.e., a feature which “[does] show stylistic stratification as well as social stratification” and “may lie below the level of conscious awareness, [but] will produce regular responses on subjective reaction tests” (Labov 1972, p. 314). There are also signs that for some speakers, Limburgian het has risen to the level of stereotype—i.e., a “socially marked form, prominently labeled by society” (Labov 1972, p. 314)—as there appear to be speakers who sometimes explicitly reject the use of het in reference to women. For example, it has been noted that “women who are not used to het as a female-indexing pronoun often find its use very reprehensible. They are bothered by Dutch het (…) and feel objectified by this description” (Bakkes 2002, p. 39; see also Hamans 1989). In such cases, the word het appears to have gained a negative social meaning, specifically one of objectification, and speakers might even consciously refrain from using it (Bakker 1992).

Social meanings are inherently more fluid and ideology-dependent than semantic and pragmatic meanings and become linked to linguistic expressions via an indexical relationship. Importantly, these indexical associations are open to constant re-evaluation, based on speakers’ ideological views and ongoing experiences, and especially in times of social change this leads to differences in interpretation of the social meaning of a given linguistic expression (Beltrama 2020; Eckert 2008; Silverstein 2003). Indexical relationships between linguistic forms and contexts exist on different levels. First-order indexicality refers to the linking of linguistic forms to sociolinguistic contexts from the perspective of an outsider; for example, the fact that het for women is a Limburgian-specific feature (i.e., het can be used for women in Limburgian but not in Dutch). Second-order indexicality indicates similar linking, but this time it happens in-group; for example, a native speaker’s intuition that while het can be used for women, there are sociopragmatic restrictions to it (i.e., the ‘age-effect’) and that the pronoun differs from its Dutch counterpart in this respect. Finally, third-order indexicality describes a situation where the linguistic form has acquired an overt additional social meaning, making the form emblematic for a certain sociolinguistic context. The objectifying connotations of het that some speakers report point to such an emblematic or stereotypical meaning. Consequently, these speakers may hold the view that it is offensive and indecent to refer to a woman with het (cf. Agha 2003, 2007; Johnstone 2016), a view which may have arisen through linguistic influence from Dutch (Bakker 1992; cf. Babel 2011), but also may be a reflection of the ideological verdict that the distinction conveys inequality between men and women, which would reflect a more global cultural impact on language use. The present study aims to document this ideological landscape through a qualitative analysis of speakers’ comments on the pronouns.

It is significant that the pronoun system we are investigating encodes the referent’s gender. Gender is a socio-culturally relevant linguistic category that is simultaneously embedded in a language’s grammar and sensitive to language users’ attitudes, the latter especially when it comes to situations where the categories which currently exist in a language are challenged. For example, the introduction of the gender-neutral third person pronoun hen ‘they’ in Swedish was initially met with many negative, even hostile reactions and strong criticism (Vergoossen et al. 2020); generally, however, attitudes normalized and improved with time, and the use of the pronoun increased as well (Gustafsson Sendén et al. 2015, 2021). In our own research on Limburgian, we also found a relationship between speakers’ general attitudes towards non-feminine pronouns, and how often they used them: speakers with a negative view towards het were less likely to produce it, whereas speakers with a positive view used it more frequently (Piepers et al. 2023). However, it is difficult to say what causes what: we do not know whether speakers use het more often because they have a more positive view of it or the other way around, or whether this is a two-way street—especially since “familiarity and exposure breeds liking” (Gustafsson Sendén et al. 2015, p. 3).

Generally, speakers of Limburgian were found to be able to accurately estimate their own pronoun use. That is, there was a statistically significant relationship between speakers’ reported pronoun use and their actual use in a production task, suggesting that speakers are, at least to some degree, aware of their own pronoun use (Piepers et al. 2023). Explicit awareness may affect the way people use a feature in non-systematic ways (Nycz 2016), guided by their personal beliefs and attitudes (Kootstra and Muysken 2019). Avoiding a feature felt to be offensive is likely to be easier when it is a content word than when it is a more grammaticalized part of speech, such as a pronoun or a grammatical pattern, given the high degree to which speakers rely on automatized production when it comes to grammar and function words (e.g., Christiansen and Chater 2016). However, the monitoring of function words and grammatical patterns is common when the socio-cultural setting triggers the salience of such elements (Hill and Hill 1980; Aikhenvald 2002; Stanford 2009; Woolard et al. 2014). If a given linguistic feature is hard to ‘get right’, and a speaker knows that ‘getting it wrong’ will have social consequences, one solution is to simply not use it (Woolard et al. 2014).

This subsection illustrated how the ideas that speakers hold about their language may influence how they use that language. In the Limburgian case, avoiding the use of the non-feminine pronoun should be relatively easy, given that a substitute—the feminine pronoun—is readily available in the language. This echoes a generalization sometimes made in contact linguistics, that changes are more often found to be ‘system-preserving’ than ‘system-altering’ (Aikhenvald 2003): it appears to be easier to increase the use of something that is already entrenched, than to introduce and propagate something new (Backus 2005).

3. Materials and Methods

3.1. Participants

A total of 155 speakers of Limburgian participated in our study (79 male, 76 female; none identified differently). They ranged in age from 19 to 83 years (M = 45.16, SD = 16.99). Participants were recruited through personal communication, social media, and a call for participants on a public regional radio broadcast. Calls for participants were additionally featured on the websites of the regional radio station and the Limburgian dialect association Veldeke.3 Participation was voluntary; ten gift cards worth EUR 15 were raffled off among the participants.

3.2. Procedure and Outline of the Questionnaire

The questionnaire was administered online through Qualtrics (2020). The introduction and instructions, as well as all the questions, were presented in Dutch, as relatively formal texts such as the instructions for and content of our questionnaire are virtually never encountered in Limburgian (cf. Schmeets and Cornips 2022). The questionnaire was preceded by an experimental task, which is not analyzed in the current paper (see Piepers et al. 2023 for details). The questionnaire itself consisted of four parts. First, participants answered questions about their personal and linguistic backgrounds. The second part contained questions about participants’ use of and attitudes towards non-feminine pronouns for women in Limburgian. The third and fourth parts of the questionnaire were collected as part of a pilot study for future research, and are not included in the current paper.

Regarding the use of non-feminine pronouns for women, we first provided participants with some brief background information. We then asked them to indicate how frequently they use het themselves, and hear others do so. Female participants were additionally asked if they themselves are referred to with het by others. Participants were further asked to answer the following open questions (note that (6) was only presented to female participants):

4.

You indicated that you {use/never use} het to refer to women. Can you tell us something about {when or why you do this/why you do not do this}?

5.

How do you feel about someone referring to a woman with het?

6.

How do you feel about someone referring to you with het?

3.3. Analysis

We subjected participants’ answers to these open questions to reflexive thematic analysis, following the procedure outlined by Braun and Clarke (2006). We used an inductive approach, meaning that we generated our themes based on the data sets. We started by familiarizing ourselves with the data set by reading and re-reading participants’ answers. Next, we generated initial codes in the data, which were subsequently collated into potential themes. These were then reviewed, adjusted, and refined so that they together formed an accurate description of our data. We generated our codes and themes both semantically and interpretatively, i.e., based on both the surface contents of participants’ answers and our interpretation of these answers (Braun and Clarke 2006, pp. 87–91).

Based on their answers, we also categorized participants’ individual general attitudes as positive, negative, or neutral. The vast majority of participants held a neutral attitude: this was true for 116 people in total (74.78% of the total; 57 female; 59 male). These participants ranged in age from 19 to 82 years (M = 45.12, SD = 16.38). The second largest group, though considerably smaller, held a negative attitude: this was true for 30 participants (19.35% of the total; 13 female; 17 male). These participants ranged in age from 20 to 83 years (M = 46.37, SD = 19.90). Finally, a small group of participants held a positive attitude; this was the case for just 9 people (5.81% of the total; 6 female; 3 male). These participants ranged in age from 20 to 60 years (M = 41.67, SD = 15.61).

In the following section, we report certain opinions and experiences which are representative of our current sample. We discuss our themes in general terms, without putting numbers or percentages on them. Where we use terms like ‘many participants’, what we mean is that something has been mentioned regularly, and by more than a few participants.

4. Awareness and Evaluation of Non-Feminine Pronouns in Limburgian

In their answers, most respondents expressed awareness of the identity of Limburgian as a separate language with its own rules, of the socio-pragmatic conditions that stimulate or constrain the use of the non-feminine pronouns, of the connotations it may have, and of the fact that different people have different opinions about it. We discuss these four dimensions in turn.

4.1. Limburgian Is Not the Same as Dutch

The most prevalent theme we identified from our data pertains to awareness of the grammar of Limburgian, and specifically how this differs from Dutch grammar. The option of referring to women with non-feminine pronouns is a traditional feature of Limburgian and a common part of everyday speech: speakers are used to hearing others around them use het, hem, det/dat, and zien in reference to women. It is an important and conventional part of Limburgian grammar, and many speakers, therefore, view it as completely normal and natural, as a fact of life:

7.

‘Seeing something as ‘good’ or ‘bad’ when it is a linguistic phenomenon makes no sense at all. I couldn’t find it ‘wrong’ that the square root of 16 is 4, could I?’

Many speakers do note that the Dutch cognate het has a different meaning, but also stress that Limburgian het by no means shares this meaning. They mention, sometimes explicitly, the contrast between the two languages, stating that they would use and/or find it appropriate in one language, but not in the other:

8.

‘In Dutch I don’t find this correct, but I do in Limburgian’.

9.

‘In Dutch I would never do this’.

10.

‘In Limburgian it’s completely normal, but in Dutch, it’s not. There it sounds strange, degrading, and unseemly’.

The use of non-feminine pronouns in Limburgian is often regarded as straightforward, and pointedly not as an act of misgendering. On the contrary: het is often characterized as “being the same as zij”. It is perceived as a ‘full-fledged’ pronoun for women, which has no insulting or demeaning overtones, and no connotations of inanimacy. The context reveals sufficient information to be able to tell whether the referent of hem or zien is a man or a woman, and it is often noted that “these words don’t carry gender”. In this sense, a contrast with Dutch is not mentioned specifically, but is instead reflected in a somewhat defensive attitude about Limburgian: nothing in our questions alluded to the pronoun use being objectifying, derogatory, or misgendering—and yet, participants felt it important to express explicitly that the pronoun use is none of these things:

11.

‘Het is, for me when I’m talking about a woman in Limburgian, not a neuter form. It’s the same as zij for me, so it’s not like I unconsciously ‘objectify’ women, or think any less of them’.

12.

‘Het means the same as zij here. It’s not just used for objects or things, like in Dutch. Using it isn’t derogatory, it really is for women, too’.

13.

‘To me, these are just words that aren’t really ‘masculine’ or ‘feminine’, so I don’t think that you misgender someone when you use het or zien for a woman’.

Moreover, the fact that Limburgian differs from Dutch in this respect is cited as a positive aspect:

14.

‘That’s just how ‘we’ say it. If I have to have an opinion about it: I like it, because it’s distinguishing from Dutch, and people always know that it’s about a woman anyway, even if the masculine form is used’.

15.

‘I’m a little torn. On the one hand, it’s outdated, and I often ‘correct’ it in a humorous way. On the other hand, there is also a certain regional charm about it’.

Note, however, that example (15) refers to the use of het as ‘outdated’, despite noting its ‘regional charm’. A similar view is echoed in (16), where the speaker mentions that while het is a normal part of Limburgian, replacing it with ziej would be a desirable sociolinguistic development:

16.

‘In Dutch I would find it misogynistic. In Limburgian, it’s part of the language, but it would be good if it also evolves into the feminine form there. In this sense, language can also follow a social development’.

4.2. Appropriateness

Many respondents showed awareness about when the non-feminine subject pronoun can and cannot be used. First, for some respondents, het and ziej are the same: both options are unmarked for them. Note that this stands in contrast to the generalizations about socio-pragmatic constraints uncovered in previous research. As a speaker, these respondents use both options interchangeably, and as a hearer, they do not care whether other speakers use het or ziej to refer to a given woman. These are speakers for whom non-feminine pronouns are unconditionally appropriate:

17.

‘It’s not for a specific woman or women. It can be a family member or a friend, but also people who you are less close to’.

For many other speakers, non-feminine pronouns are only conditionally acceptable and appropriate. Whether this is the case, is dependent on various factors.

Participants frequently discuss a woman’s age as a guiding factor for the choice of pronoun. For younger women, het is reportedly used more often, but for older women, feminine pronouns are often preferred. Two related factors, personal closeness and degree of respect, are also mentioned frequently. When speakers are closer to the woman they refer to, they can use het, but when the personal distance is larger, ziej is preferred. Similarly, when a speaker’s degree of respect for a certain woman is high, ziej is preferred. These three factors can conflict with one another: for example, grandmothers, even though speakers are close to them, are generally not referred to with het, due to their higher age, and the high degree of respect speakers have for them. In this sense, age and respect appear more important than personal closeness in determining the appropriateness of het.

Speakers qualify ziej as the formal (or ‘polite’) third-person pronoun, which goes hand in hand with the formal second person pronoun (Limburgian has a T/V distinction in second person) as the appropriate term of address. That is, if a woman is addressed with the formal second person pronoun, she is referred to with ziej in the third person. Note, however, that this is not necessarily true the other way around: a person can be addressed with the informal pronoun, and still be referred to in the third person with ziej (Bakker 1992). Moreover, as example (18) illustrates, the woman’s age should be interpreted as relative to the speaker’s age:

18.

‘You do that with people you know very well, a sister or a cousin, for example. Aunts and people from an older generation are addressed with the polite form: geer ‘you(formal)’, and ziej and not het. It’s clearly something for younger women in general. Of a sister, it can be said for life’.

Speakers reported in general that it depends on who they are talking about whether they will use the non-feminine pronoun. Accordingly, het is acceptable or not for some participants:

19.

‘It is usually about someone the speaker knows. Intuitively, that feels right. But if it is about a stranger, for example, then it sounds strange’.

20.

‘About a woman I find this quite indecent, about a girl I find it normal’.

One woman specifically mentioned that she does not mind being referred to with het, as long as she is on a first-name basis with the speaker. Similarly, the formality of the context is often cited, with het being appropriate in informal conversations.

21.

‘No problem at all, if they normally address me with my first name’.

Speakers also report sensitivity to their interlocutors as a factor. In many cases, they report mirroring the person they are speaking to, and when that person uses non-feminine pronouns, they will use them, too. The speaker in (22), a 32-year-old woman, mentions that the use of het by others helps her use it, too, as ‘her own Limburgian’ is closer to Dutch:

22.

‘It is contagious when someone else already uses those words, because my Limburgian is so ‘Dutchified’ that I now tend to use sentences with zie’.

Example (23) is from a 54-year-old male. Interestingly for someone from this generation, he was taught not to use het because in his mother’s family, it was allegedly seen as rude—which casts doubt on the idea that a negative attitude towards het is a completely new phenomenon. The speaker nevertheless uses it himself, in largely the same way as many other speakers. He also notes that in his view, it is ‘better Limburgian’, echoing the perception discussed in Section 4.1 that this is a traditional, distinctive dialect feature.

23.

‘I do that mostly when my conversation partner does it. I feel like it is ‘better’ Limburgian, but I wasn’t taught this at home, because in my mother’s family, a generation earlier, they thought it was rude, so my mother never said het and hem about women. I do know it from my peers at school. It sounds very familiar, but not from my own family. I would use het mostly for women I know well and/or are younger than me’.

4.3. Connotations

As the quotes above already intimated, participants’ characterizations of the non-feminine pronoun go beyond semantics. The use of het can give rise to inferences about the speaker, the referent, or the relationship between them. As to be expected, given that there are speakers who claim never to use het because they dislike the pronoun, there are women who do not like being referred to with het, as it makes them feel objectified. Further negative views expressed about the non-feminine pronoun were that it sounds anti-social, impolite, degrading, or even ‘wrong’:

24.

‘I don’t think this is appropriate’.

25.

‘It’s not nice, it feels degrading’.

26.

‘It’s not very nice. It suggests that a woman is an object’.

27.

‘When someone refers to a woman with het, it sounds wrong, and I interpret this as slightly degrading or disrespectful’.

Conversely, however, there were also participants who voiced positive views. Positive connotations that participants mentioned include youthfulness, seeing the referent as ‘cute’ or ‘sweet’, and the warm familiarity of a Limburgian-specific feature.

28.

‘It makes me feel like I’m still young. So that’s not unpleasant’.

29.

‘It sounds very familiar’.

4.4. Navigating Differences between Speakers

Many speakers show they are aware that the pronoun may trigger judgments from other speakers, and that it is a potential social minefield. The difference between (30) and (31) is especially striking. Example (30) was written by a speaker who referred to her mother with het and was promptly told by a hearer that she was being disrespectful. As a result, she stopped using het:

30.

‘A few years ago, I referred to a person with het, until I was roughly reprimanded one time, then I unlearned it. This happened in the context where I was talking about my mother, by saying ‘het werkt bie ut AZC’ (‘she works at the asylum center’). They said this was very disrespectful towards my mother, and in that moment, my decency seemed to have disappeared. That was of course not my intention, so I unlearned it’.

Note, however, that we cannot be sure of the hearer’s intention here: while the speaker appears to have interpreted their comment as het being rude in general, the hearer may well have just meant that het should not be used for one’s mother, specifically (cf. Section 4.2). In contrast, (31) is written from the perspective of a hearer who heard someone refer to his grandmother with het. Rather than interpreting this as rude, this participant found it sweet how the speaker talked about his grandmother:

31.

‘It’s normal, it’s a sign that the speaker sees her as youthful. My friend once used het to talk about his grandmother, who had just passed away. That made me picture that grandmother like a girl. I think het is a sweet form, it’s not at all degrading or anything’.

The effect het has on a hearer may depend on the attitude which the hearer expects the speaker to hold. For example, if a hearer (who may or may not also be the referent) dislikes het but knows (or assumes) that the speaker frequently uses it with no bad intentions, they will also know not to disapprove of the speaker’s use of het. Even though they personally do not like the pronoun, they know that others see nothing wrong with it, and therefore also know that it is probably not the intention of the speaker to be rude (as opposed to example (30), where the hearer did not take the speakers’ intention into consideration). This is illustrated below:

32.

‘Actually, I hate het or det for a woman: for me it has a negative connotation, even though I found out that that’s not always the speaker’s intention’.

Conversely, if a hearer takes no issue with het but expects that the speaker’s intention is to be rude, they can dislike the pronoun in this context even though they usually would not mind it. That is, het will be interpreted as derogatory by this type of hearer if it is thought to be intended that way by the speaker. A speaker’s perceived intention thus appears to play an important role. The consensus is that non-feminine pronouns are acceptable as long as they are not used with bad or insulting intentions.

33.

‘I’m okay with it. Only when it is intended negatively, for example, ‘oh, het again!’. That hits me harder than if the feminine form had been used’.

34.

‘Basically, it’s no problem, it’s grammar. Tone and context are way more important: when that’s negative, I don’t agree with the use, and then it’s degrading’.

35.

‘Being referred to with het doesn’t bother me, but it could bother me when the person suggests something negative with it, for example through intonation’.

36.

‘In a negative context I’d experience it as unpleasant, but as a joke, or in a normal sentence, I won’t find it bothersome’.

Finally, speakers who perceive het as normal sometimes report that they understand the point of view of those who do not like the pronoun:

37.

‘I find it normal and don’t see it as negative, but I have often heard that people find it strange or even degrading. I can imagine that, too’.

This shows that some language users are, in fact, aware that there are different intuitions and attitudes regarding het, which becomes even more evident when we look at participants’ descriptions of the conversations in which they use the pronoun. Some might know that they have to ‘navigate’ along different attitudes depending on the context and the conversation. For example, (38) shows one speaker reporting to use het only in conversations with people who are from the same geographical area:

38.

‘I use it when I speak with someone who is also from the south. When I speak with people from the middle and the north of Limburg, I usually leave it out, because I know that it is not used there. In any case, I tend to switch to a kind of regiolect when I speak with people from outside the Heuvelland’.

This speaker, a 48-year-old man from the south of Limburg, assumes that het is a feature specific to the south, and therefore refrains from using het in conversations with speakers from the middle and northern parts of the province. In his mind, het is not a part of the dialects spoken there, which, actually, it is: het and similar pronouns are used by speakers from all over the province of Limburg, with only small exceptions. This example illustrates the relevance of folk linguistic ideas, because “regardless of how misguided or unfounded [these] ideas are, they nevertheless have the power to shape behaviors and attitudes” (Miller 2021, p. 3).

Although function words are usually produced without much conscious planning (Christiansen and Chater 2016), many of the examples we provided in this section suggest a relatively high level of metalinguistic awareness, at least in a subset of participants. Speakers are not only aware of the feature itself and of their own use of it but they also know that there may be differences—regional or individual—in how this feature is manifested across the language, which may cause them to tread lightly when referring to women. This is not without reason: the examples in this section showed that while some language users set aside their own personal negative attitudes in conversation with ‘het-users’, others might not, suggesting that het might be enregistered (Agha 2007) for at least some language users. The speakers in (39) and (40), for example, explicitly link the use of het to a certain social class (note that (40) refers to the pronoun use as an ‘accent’).

39.

‘I think it’s degrading; it sounds antisocial, I also link this language use to the bad neighborhoods in Roermond’.

40.

‘I think this is a typical vulgar accent from Venlo’.

4.5. Absence of Awareness

The previous subsections have shown that many speakers have a relatively high level of awareness regarding these pronouns. As to be expected, however, this was not true for all speakers; instead, some of them noted that they were unable to answer our questions. This is illustrated in the following examples:

41.

‘Difficult question. I’m not aware of my choice of words’.

42.

‘I don’t know, it happens by feel’.

43.

‘No idea. I think I use it interchangeably, depending a bit on the person I’m speaking to. I mirror my language. I’m not language conscious, and I don’t really have a good sense of language’.

5. Discussion

5.1. Sociopragmatic Gender Is Subject to Opinion

Our results confirm that speakers are aware of the socio-pragmatic conditioning of non-feminine pronouns (i.e., het) for female referents in Limburgian (Piepers et al. 2021, 2023), specifically the previously found tendency for the selection of these pronouns to be dependent on the referent’s (relative) age, personal closeness to the speaker, and respect for the referent, like in various other Germanic varieties (Busley and Nübling 2021; Martin 2019). These pragmatic conditions, however, do not translate into a solid rule that holds across all language use for all Limburgian speakers. First, speakers can always opt for the feminine pronoun ziej, and some speakers indeed do, even in situations where other speakers would deem het the (more) appropriate form. Second, speakers differ in the extent to which they feel that there is, in fact, a pragmatic rule: for some, het is only an option in certain circumstances depending on e.g., age, closeness, and informality; whereas for others, het is always appropriate regardless of the referent or situation.

Many speakers were able to reason about their own use of het and its acceptability and provided detailed accounts of their use of het in general or in specific situations. Our findings thus demonstrate a high level of metalinguistic awareness regarding the pronouns in many of our participants. This high awareness, however, does not usually translate into relatively outspoken positive or negative attitudes: most speakers reported simply seeing the pronoun as part of the language. However, there are also speakers who engage with the pronouns more actively, either criticizing or defending their use. Moreover, some speakers indicated that they knew that pronoun choice may be an issue in the community, even if they do not have strong feelings about it themselves. Based on this, we conclude that pronoun choice is salient to some degree, which makes it likely that for many the choice of how to refer to a female referent in Limburgian requires conscious attention and sometimes active decision making. This is further supported by the finding that participants’ attitudes towards the pronoun also correlated with their actual use (Piepers et al. 2023).

5.2. The Role of Dutch

The participants in our study often commented on the fact that Limburgian speakers also speak Dutch, on the differences between the two languages regarding pronoun use, and sometimes on how they view this bilingual reality. The substantial variation we found between speakers indicates an instability in the system, which in turn may be indicative of an ongoing change (Backus 2014; Croft 2000). If we assume that our data do reflect a change in progress, an obvious potential source is transfer from Dutch. Speakers are bilingual in Limburgian and Dutch; Dutch plays a dominant role in many people’s lives as the language of, e.g., education, work, and media; it is also often used with family and friends (Schmeets and Cornips 2022). Dutch, as mentioned, does not have an equivalent use of the neuter pronoun het as a subject pronoun for female referents. In such circumstances of sociolinguistic status asymmetry and structural difference, contact-induced change is common (Croft 2000; Thomason 2001), and it is likely that the Limburgian dialects undergo massive impact from Dutch (cf. Giesbers 1986; Vousten 1995). The non-use of the neuter subject pronoun might then straightforwardly be seen as influence from Dutch, where it encodes that the referent is inanimate. Since this meaning is exceptionless in Dutch, and therefore may be assumed to be maximally entrenched for Dutch speakers, it may be expected that in the minds of Limburgian speakers, this entrenched Dutch unit influences the cognate Limburgian one, especially since the latter is sensitive to relatively subtle pragmatically conditioned variation, as discussed in the previous sub-section.

However, the question remains as to why that influence appears to affect speakers so differently in this case. If it was merely a matter of Dutch influence on the frequency of use, one would expect a general decrease in the use of the pronoun, but this is not what we find. The pragmatic conditioning visible in both the way the pronoun is used and in what people say about it, suggests a degree of stability, as most participants use the pronoun and most have a neutral attitude about it. The only clear change is that some people simply do not use the pronoun at all, and sometimes cite their ideological reasons for it—on the other hand, there are others who cite ideological reasons for preserving it (“Dutch and Limburgian are not the same”). Our data suggest that pronoun use is sensitive to differences across the population in the degree people are aware of it and in the attitudes they hold about it.

Cross-language activation, to borrow the words of Kootstra and Muysken again, is “a prime example of language contact in the bilingual mind” (Kootstra and Muysken 2019, p. 1). Our suggestion is that the semantics of Dutch het meeting those of Limburgian het in the mind of a single speaker can have, broadly speaking, two effects: a peaceful merge, or a clash. Many speakers maintain a neutral attitude towards sociopragmatic pronouns, often stating something along the lines of “that’s just how the language works”, suggesting these speakers experience peaceful merging: they do not notice, do not care, or both. The speakers with negative and positive attitudes, however, have experienced a clash: they do notice the differences, and, importantly, they do care. For some speakers, the Dutch semantics have been borrowed into Limburgian, leading them to overtly negative attitudes and, in a sense, “complying with” the dominant language. For others, the difference appears to be a reason to double down on the original meaning and thus, in a sense, “rebel against” the dominant language. The following excerpts from previous examples illustrate this point:

44.

‘My Limburgian is so ‘Dutchified’ that I now tend to use sentences with zie’.

45.

‘I like it, because it’s distinguishing from Dutch’.

46.

‘It suggests that a woman is an object’.

It appears that our object of study, i.e., pronouns that ‘mismatch’ in the two languages of a bilingual in such a way that in one it is completely acceptable and in the other its cognate is not, is situated at the crossroads of the cognitive and the social. The same linguistic entity, the non-feminine subject pronoun het in Limburgian, may be used or not used by individual speakers, and its use or non-use may be conditioned by cognitive automaticity or by a conscious, emotionally charged, consideration about its desirability. The relevant social driver of language use is that speakers may find their traditional language important and cherish that it is different from the standard language, or, alternatively, may desire to avoid insulting anyone and avoid using a pronoun that might raise some eyebrows. The relevant cognitive driver is that, as a result of using and experiencing a huge amount of Dutch, speakers may find themselves replacing het by ziej more often, a case of unintentional transfer; cf. (44). While it is clear that speakers can attend to any linguistic element and make it the focus of conscious selection (or avoidance), it is equally clear that much of language use proceeds more or less on automatic pilot. Whatever the ultimate reasons for individual choices, if the rate of usage of the different pronouns differs across individuals, they may also have the different pronouns entrenched to different degrees, which could lead to differences across speakers in the ease with which they activate the different pronouns, potentially affecting further pronominal choices (Backus 2014; Bybee 2010; Dąbrowska 2012, 2020). Language, as Dąbrowska (2020, p. 224) puts it, “is learned from examples of utterances produced by speakers engaged in communicative interaction”, and “can live only in individual minds”. Speakers may make different choices in pronominal reference due to differences in entrenchment of the individual forms, not just because of differences in attitudes that affect whether or not a speaker finds it appropriate to use the non-feminine gender in specific communicative contexts. When examining speaker agency, as we do in this paper, it is important to not lose sight of the pervasive impact of automaticity, as it has the effect of limiting speaker agency.

The result is variation that is at least partially sociolinguistically conditioned but also relatively unpredictable because people’s attitudes differ and people also differ in the degree to which their attitudes influence their speech. This highlights why the integration of social and cognitive triggers of language use is of vital importance for understanding language use (e.g., Geeraerts et al. 2010). Communication happens in a socio-cultural context but also relies on a cognitive architecture that promotes automated activation of highly entrenched units. Psycholinguistic methods are, of course, well-suited for investigating the mechanisms and causes of automatic production, and sociolinguistic methods are equally well-suited to investigate agency in language use. The challenge is to design studies that integrate these two perspectives and methodologies.

5.3. Ideologies in a Non-Standardized Language: A Breeding Ground for Enregisterment?

An important aspect of Limburgian-Dutch bilingualism is that there is a significant status inequality typical for contexts widely perceived to be cases of dialect-standard language co-existence. Limburgian is not a standardized language, and its written tradition is mostly limited to typical dialect literature (e.g., poetry or stories in the local language variety). There are, therefore, no grammar books or education systems that explain when het is appropriate and when ziej should be used, and no classrooms in which such rules would be learned and practiced. Without a regime of authority, variants are less easily criticized as ‘wrong’ or ‘uncivilized’. The more an element is seen as simply triggered by ‘grammar’, as part of the linguistic system, the more its selection may be assumed to be under the radar. However, when two forms are possible for the same meaning, and their selection is dependent on a judgment call regarding sociopragmatic suitability, it stands to reason that their selection in running speech may not always be completely automatic.

Once speakers are aware of the social connotations of a form, they may start consciously deciding whether or not to use it. The Limburgian pronoun issue seems to fit the bill. If speakers feel that they should not use the pronoun, and subsequently try to avoid its use, they engage in grassroots linguistic engineering (as opposed to the kind where an authoritative body attempts to regulate how people speak through official language policy; cf. Kristiansen and Coupland 2011). Pronouns are furthermore an interesting case: on the one hand, they are clearly caught up in societal discussions about how to refer to underprivileged groups, and therefore they are prime targets for conscious attention, just like, e.g., nouns that designate social groups. On the other hand, and unlike such nouns, pronouns are function words that are part of grammatical templates. Such templates are often produced ‘under the radar’, presumably because high entrenchment levels facilitate automatic activation. Subject pronouns have, for example, been shown to be sensitive to priming (e.g., Travis et al. 2017), typically a phenomenon that relies on automatic production rather than conscious selection.

The question remains where we should situate the Limburgian pronoun choice in the classic sociolinguistic distinction between markers, indicators, and stereotypes (Labov 1972), and whether it should be seen as indexing a particular way of speaking and its associated identity, and therefore be enregistered (Agha 2003, 2007; Johnstone 2016). Our findings suggest the stirrings of a grassroots movement that discourages the use of these pronouns; at this point, it is impossible to know whether more and more speakers will join the currently small group of people who seem to be actively avoiding the non-feminine pronouns, or the equally small group that actively promotes its use. One can imagine that increasing polarization of opinions about pronoun choice could lead to a situation in which every pronominal choice becomes a political statement: use indicates an active stance in favor of Limburgian tradition and non-use indicates an active stance against sexism. At this point, the data do not give us any indication to think that Limburgian society has arrived at this situation or will arrive at it at any point. What we can say, however, is that for some speakers, pronoun choice has become—or is becoming—an indicator of an identity. That means the seeds are there for the future development of pronoun choice for female referents in Limburgian into a widely recognized index of socio-cultural orientation. Future research could seek to correlate speakers’ ideological stances and particular communicative constellations of interlocutors with the use of non-use of the non-feminine pronoun.

6. Conclusions

We conclude that we may be witnessing the early stages of an ongoing language change in Limburgian, which takes its specific shape in the context of bilingualism with Dutch and a socio-cultural movement promoting gender equality. Non-use of the neuter subject pronoun to refer to women, a traditional aspect of Limburgian grammar, may be due to influence of Dutch, as all the Limburgian speakers use this language extensively every day. At the same time, it may also be the result of a growing sense among part of the population that the pronoun raises awkward connotations of inanimacy or objectification. The attitudinal data discussed in this paper show that many speakers are aware of the issues involved, though their responses to them are far from uniform. Some are positive about the continued use of the non-feminine pronoun, others are negative, and the majority hold neutral attitudes. Pronoun choice might develop further into a social issue within the speech community if more speakers start seeing the gender mismatch not as just a neutral feature of Limburgian grammar, but as an expression of sexist vocabulary and subliminal encoding of gender inequality. Whichever way Limburgian pronoun use develops, the case is interesting for the difficulty involved in establishing the degrees to which it is caused by language contact and by larger socio-cultural or ideological trends.