Introduction

Language attitudes (LAs) have a long history of influencing language use, language variation and change, language shift and language maintenance (Garrett, 2010; Kircher and Zipp, 2022; Sallabank, 2013). Therefore, understanding LAs may help researchers identify positive or negative patterns in attitudes toward languages or regional varieties, and researchers can recommend suggestions based on experimental evidence. Overall, research has revealed that LAs operate similarly around wider attitudinal dimensions, for example, status vs. solidarity (see El-Dash and Busnardo, 2001; Genesee and Holobow, 1989), status, solidarity and dynamism (see Zahn and Hopper, 1985) and status, solidarity and cognitive development (see (Kircher et al., 2022). Attitudinal dimensions may reveal how people perceive and evaluate different languages or language varieties. The question now is whether the widely revealed attitudinal dimensions in previous research hold true in fast-growing societies. In other words, can we gain new insights into the attitudinal dimensions in these countries? This paper will examine LAs in a fast-growing society and reveal how LAs are driven in such societies.

In recent years, many rapidly developing countries in the Middle East have seen tremendous growth and advancement. Among them is Saudi Arabia (Mati and Rehman, n.d.). Saudi Arabia is a large country in Western Asia with a population of nearly 35,000,000 people (General Authority for Statistics, n.d.). Saudi Arabia has gone through rapid developmental changes in all sectors since the discovery of oil in 1938. Furthermore, recently, in April 2016, Crown Prince Mohammed bin Salman announced an ambitious, extensive modernisation project under the Vision 2030 framework in the country. The Vision’s fundamental goals are ‘to diversify the country’s economy while also implementing significant social and cultural reforms’ (Habibi, 2019, p. 1). The Vision has achieved many goals so far, including social and cultural reforms, which could have their own implications for the sociolinguistic situation in the country.

Linguistically for Saudis, Arabic is the native language. Many regional varieties of Arabic are spoken in Saudi Arabia. These dialects are diverse and geographically bounded (Aldarsoni, 2011; Ingham, 1994; Prochazka, 1988). In this regard, recent research has mostly investigated Saudi dialects from variationist sociolinguistic perspectives to investigate how variations between Saudi dialects are structured and discover the factors for such variations (see Alaodini, 2019; Al-Essa, 2009). However, research examining LAs and language ideology toward Saudi dialects has been mostly overlooked.

Given the significant social and cultural reforms in Saudi Arabia brought about by urbanisation, the present research is especially timely as it examines LAs in Saudi Arabia at a time of rapid social change. This paper will ask two research questions: first, what are the attitudinal dimensions of Saudi Arabian LAs, and how are Saudi dialects associated with the attitudinal dimensions? Second, what are the effects of socio-demographic variables on the dimensions of Saudi Arabian LAs?

The study uses the verbal guise technique as presented by four male speakers representing four salient dialects in the country (i.e., Najdi, eastern, northern and southern dialects). A total of 568 Saudi participants with diverse demographic backgrounds were recruited for the study. The data were analysed using factor analysis (FA) and ANOVA.

Based on the findings, the study aims to contribute to the advancement of the theory of LA research by challenging the dominance and significance of the dynamism dimension in rapidly urbanised countries, such as Saudi Arabia. Furthermore, the study aims to reveal the implications of social background on Saudis’ attitudes toward media and education.

The findings are subject to three key limitations: first, the speakers were all men. Although every effort was made to obtain a proportional number of men and women, this was unsuccessful. Second, the sample recruited for the study is 568 Saudis, which is statistically less (see section “Participants”) than the required sample size for a more generalisable study. Third, the Hijazi dialect has been excluded from the study as the Hijaz region has a complex linguistic and social situation different from all other regions in Saudi Arabia. It is better for research to regionally design a method suitable for this complex situation. The researcher would have to base her investigation on the social groups living in the region rather than dealing with the whole region as one that represents a homogenous social and linguistic group. Thus, in the current study, I preferred to exclude Hijazi to avoid negative effects on the validity of the study.

LAs and language ideology

Attitude structure contains three parts: cognition, affect and behaviour (Kristiansen et al., 2005). Cognition is responsible for the beliefs that a person has regarding the world around them (Garrett, 2010). Most importantly, cognition involves the associations that a person makes between objects around them (Cargile et al., 1994). For example, the association between a particular social group and a particular dialect is viewed as coming from cognition. The effect involves how positive or negative one’s feelings are towards an object (Kristiansen et al., 2005), and behaviour is the direction of the attitude itself (Kristiansen et al., 2005). Regarding the aim of the present paper, cognition is highly significant in interpreting this study’s results because this paper uncovers the dimensional model behind Saudis’ attitudes toward their dialects. Once the study taps into cognition, it brings a better understanding of the associations that participants make between Saudi dialects and their associated characteristics.

Language ideology is the wider system of beliefs that maintains and triggers LAs (Garrett, 2010). Fuller (2018) emphasises that the difference between the two concepts lies in the methodological approaches in research. LA research investigates views and beliefs held by speakers through traditional LA methods (i.e., direct, indirect and societal treatment), while language ideology uses discourse analysis methods to uncover the wider macro system of beliefs. Thus, it is argued that ‘Language attitudes and ideologies clearly interact and influence each other, and the lines between them may become blurred’ (Fuller, 2018, p. 121, 122).

LA research

Based on a large body of research, LAs influence the ways individuals perceive language. What follows is the interpretation of such perceptions into the categorisation and stereotyping of languages and varieties. Such stereotypes might range from positive to negative. Positive stereotypes would favour some languages over others, while negative stereotypes could drive people to practise bias and discrimination when language is used for communication (Garrett, 2010; Garrett et al., 2003; Sallabank, 2013). Such negative practices are essential to be understood by sociolinguistics as a way to suggest solutions for these problems. For example, LAs play a significant role in the legal context; a study by Seggie (1983) revealed how people classify the degrees of crimes according to the speakers’ accents. In a multilingual context, LAs reveal how LAs could be a driving force in determining whether or not to pass a specific language down to children (Kircher et al., 2022). In the search for housing, LAs reveals how attitudes towards the accents of potential tenants over the phone could be a determinant for a landlord to rent the individual a house or not (Purnell et al., 1999). In an endangered language context, many studies have shown that negative attitudes towards an endangered language might increase the chances of its loss; similarly, positive attitudes might increase the chances for the language to be maintained and revitalised (Sallabank, 2013). Furthermore, research on LAs towards endangered languages has helped language planners to propose projects for language revitalisation (Kircher and Fox, 2019; Yin and Li, 2021).

LA research has a powerful ability to uncover peoples’ overt and covert attitudes towards dialects and languages (cf. Garret et al., 2005a, 2005b; Giles, 1970; Ladegaard, 1998) as it taps into subjective factors (i.e., attitudes) through objective research methods. Therefore, the choice of research methods in studies on LA plays a significant role in eliciting attitudes. Research has proven that overt attitudes are usually uncovered through a conceptual approach, in which the name of the dialect is explicitly revealed, whereas covert attitudes are usually revealed when the dialect is represented through a voice sample, and the participants are asked to rate the speaker rather than the dialect itself. Lambert et al. (1960) used voice samples to elicit participants’ covert attitudes using the matched guise technique (MGT). In the MGT, voice samples are presented by one speaker who mimics all the dialects, and the participants are asked to rate the voice of the speaker rather than the dialect itself. An alternative to the MGT is called the verbal guise technique (VGT). Although it follows the same procedures as the MGT, each dialect is represented by its native speaker (Garrett, 2010). The present research uses the VGT for several reasons. First, it will elicit people’s attitudes indirectly. If participants are aware of the purpose of the research (i.e., conceptual approach), they might avoid providing real and authentic attitudes and instead provide attitudes that they think would please the researcher (Garrett, 2010). Second, the VGT was chosen over the MGT as it is difficult to find a speaker who can accurately mimic other dialects (Garrett, 2010).

It is important to note that LA research methods inherently have drawbacks. Preston (1989, 1999) argues that when participants listen to the dialect of a speaker, it is difficult for the researcher to be certain that the dialect that is presented to them is the same dialect that the researcher thinks of cognitively. Therefore, he proposes asking participants to identify the original regions of the varieties that are presented to them, either by allocating the dialects to their original places on a given table or placing them on their geographical locations on a map. As a result, the researcher would be confident that the presented dialect is identical to the dialect that the participant thinks of and rates accordingly. Many studies have integrated the language identification approach into LA research (see Bayard et al., 2001; Dragojevic et al., 2018; McKenzie, 2008, 2015). Therefore, the present study will include the dialect recognition question in the questionnaire to ensure that the dialect that the researcher thinks of aligns with the dialect that the participant rates.

Many global and cross-cultural studies on LAs have found that two attitudinal dimensions are always revealed: status (i.e., prestige) and solidarity (i.e., social attractiveness) (see El-Dash and Busnardo, 2001; Genesee and Holobow, 1989). The status evaluative dimension is associated with perceiving the investigated language as prestigious, correct and standard, whereas the solidarity dimension is associated with the language being perceived as pleasant, friendly and socially attractive. It is important to note that research has revealed that two patterns are associated with these dimensions. The first is that standard varieties load higher in the status dimension than in the solidarity dimension (see BBC, 2005; Giles, 1970; Ladegaard, 1998). The second pattern indicates that the participants of the dialects that are rated high in solidarity have in-group loyalty (Dragojevic et al., 2013).

Zahn and Hopper (1985) revealed that a third evaluative dimension was manifested in their study, the dynamism dimension. This refers to language being ‘for example, energetic/lazy, enthusiastic/hesitant’ (Garrett, 2010, p. 55). Furthermore, Kristiansen (2001) found that the widely revealed status-versus-solidarity dimensions were not significant to Danish participants compared to superiority (i.e. status) versus dynamism. Moreover, Kristiansen (2009) found that dynamism was much more important than superiority to Danish participants due to the emergence of the neo-standard in the country. This result might suggest that the strength and presence of the dynamism dimension are driven by social and linguistic changes. Therefore, given the rapid urbanisation changes in Saudi Arabia, the present study is timely in determining whether or not dynamism is driven by change and social reforms.

LA research in Saudi Arabia

In Saudi Arabia, LA research is a growing field that mostly focuses on investigating attitudinal reactions towards regional dialects rather than exploring all dialects in a broad and national approach, or it addresses attitudes from the point of view of one social group towards other dialects in Saudi Arabia. For example, Alrumaih (2002) only investigated attitudes towards the main varieties in Saudi Arabic from the point of view of 60 Najdi participants (i.e., residents of the central region). Furthermore, Alrumaih collected attitudes towards Saudi Arabic dialects by only including two characteristics in his questionnaire, taken from previous studies: correct and pleasant. This might open a line of questioning about whether these characteristics reflect real evaluative profiles of Saudi dialects. There might have been more suitable characteristics revealed that could precisely reflect LA in Saudi Arabia if a piloted study has been conducted to elicit more characteristics. The results of Alrumaih’s study revealed that Najdi was rated as the most correct and pleasant dialect, southern as the least pleasant and eastern as the least correct.

Similarly, Aldosaree, (2016) examined LAs towards only three dialects in Saudi Arabia: the Najdi dialect, the Western dialect and the Southern dialect. The sample was 66 college students and 69 high school students. He used the VGT, and he interviewed 12 participants. The main research question focussed on the possibility of having different attitudes when educational levels vary. The results of the study revealed that high school students had a tendency to express negative attitudes about speakers of out-group dialects, while college students had less biased attitudes. Aldosaree’s study provides significant findings in the field of LA research in Saudi Arabia, but the study was confined to only three dialects, making it unable to provide a broad picture of LA in Saudi Arabia. In addition, the adjectives provided in the scale did not run through a piloting study, which might have affected the reliability of the results. Rather, they were chosen anecdotally because they are common in Saudi Arabia. Finally, the analysis of attitudes relied on counts and percentages, lacking an exploration of the dimensions behind them.

Other researchers in Saudi Arabia have investigated LA towards sub-dialects (see Alahmadi (2016) for LA towards Meccan dialects, a sub-dialect of the Hijazi dialect, and Al-Rojaie (2020a) for attitudes towards the Qasimi dialect, a sub-dialect of the Najdi dialect).

Alhazmi (2018) investigated LAs in Saudi Arabia based on FA in only one region in the country, the Hijazi region (i.e., Hijaz). The findings of her study revealed a new and original dimensional model that includes two dichotomous dimensions, modern and traditional. Alhazmi claims that the novelty of the dimensions in her study is due to the complex social situation found in Hijaz. The region receives an annual influx of pilgrims every year. Some visitors stay in the region as Saudi Arabia has offered increased employment opportunities since the discovery of oil in 1938. Thus, the advent of external migration to Hijaz has massively affected the linguistic situation and resulted in a dichotomy in which the two main dialects spoken carry different linguistic features from each other, one dialect representing the original Bedouin society and the other representing the migrated Hadari (sedentary) society. No other region in Saudi Arabia has a similar linguistic situation to Hijaz: other regions have one broad dialect and corresponding sub-dialects, while the linguistic situation in Hijaz is notably dichotomous (Alhazmi 2018).

Al-Rojaie (2021) examined perceptual dialectology and the LAs of 674 Saudis toward dialects in Saudi Arabia. Al-Rojaie used two techniques from perceptual dialectology methods: a map-drawing task and a labelling task. Participants were provided with a map of Saudi Arabia and were asked to draw and label lines/boundaries around dialect areas in the country. As for the label task, which was mainly focussed on LAs, participants were asked to give a labelled characteristic for each dialect area they had drawn. The study only presented labels that were relevant to the study as some labels were not attitudinal in nature (e.g., alternation of vowels, influence, vocabulary words, and social media categories). The study revealed that first, the Najdi dialect was mostly associated with labels such as ‘clear and aggressive’. Second, the southern dialect was mostly associated with labels such as ‘unclear, closed, and heavily influenced by Yemeni dialect’. Third, the eastern dialect was highly associated with ‘vowel lengthening and was mostly influenced by the Gulf dialects’. Fourth, the Hijazi dialect was highly associated with labels such as ‘light, soft, open, urban, peculiar vocabulary, mixed dialects, and attractive’. Finally, the northern dialect was the least labelled with characteristics; two of the most frequent that appeared in the northern dialect were ‘Bedouin’ and ‘attractive’ (Al-Rojaie, 2021, p. 512, 513).

In a similar vein, Alfalig and Alhazmi (2022) have adopted a national approach to explore Saudi attitudes towards the main five dialects in the region. Their study was a preliminary work that has the potential to pave the way for further studies, such as the current one. They used keyword techniques to collect attitudinal data. They asked 78 participants to jot down the five main characteristics of the dialects that were presented to them conceptually in the questionnaire. Their research revealed that, first, Najdi was characterised as ‘arrogant’, ‘easy to understand’ and ‘prestigious’. Second, the southern dialect was highly described as ‘the most difficult to understand’, ‘fast, harsh and uncultured’. Third, the eastern dialect was categorised as ‘slow’, ‘heavy accented’ and ‘quiet’. Fourth, the Western dialect was described as ‘friendly’ and ‘easy to understand’. Finally, the northern dialect was frequently characterised as ‘generous’, ‘moralistic’ and ‘Bedouin’ (Alhazmi and Alfalig, 2022, p. 117, 118). Building on the results of this research, the present study takes advantage of FA to reveal the latent dimensions behind LAs in Saudi Arabia.

Although Al-Rojaie (2021) and Alfalig and Alhazmi (2022) attempted to investigate LAs on a national level, their studies did not delve further to reveal the wider attitudinal dimensions behind Saudi dialects and how these dimensions may be affected by demographic variables. Thus, the present study will fill in this gap in the literature by examining the attitudinal dimensions of LAs in Saudi Arabia.

All previous research in the region has been introductory, uncovering a vivid but fragmented picture of LAs in Saudi Arabia. Therefore, the present study will adopt a broad approach in which the main dialects of Saudi Arabia are included in the study to uncover a holistic attitudinal dimensional model that can be used as groundwork for any future attitudinal studies in Saudi Arabia. Since the aim of the present study is to focus on the remaining unexplored regional dialects in Saudi Arabia, the western dialect (i.e., the Hijazi dialect) will be excluded from the present study. This is because the region has a complex social situation that requires any attitudinal study to investigate it separately from other Saudi Arabian dialects (see section “Limitations” for more details on the linguistic situation in Hijaz).

Saudi Arabian dialects: classifications and geographical locations

Arabic is spoken widely in the Middle East and some parts of northern Africa. In total, 23 countries speak Arabic. In this respect, it is important to note that a diglossic situation exists in Arabic countries in which two types of Arabic operate in two different settings: modern standard Arabic (MSA), and dialectical Arabic, such as Saudi Arabic or Jordanian Arabic (Alghamdi, 1998; Ryding, 2005). The former is regarded as the high variety and the most correct as it is used in formal situations such as education, official functions and religious speeches. The latter is viewed as the low variety since it is used in informal situations (Marley, 2004). MSA is viewed as the most correct variety as it is the closest to the language of the holy Qur’an, which is in classical Arabic. Alotaibi and Borsley (2013, p. 22) argue that ‘an important fact about MSA is that it is not anyone’s native language’. Therefore, the present study will deal only with dialectical Arabic in Saudi Arabia as MSA is not a language that is native to Saudis.

Regarding the dialects spoken in Saudi Arabia, previous research has revealed that they are diverse and geographically bounded (Aldarsoni, 2011; Ingham, 1994; Prochazka, 1988). Many researchers have investigated the classifications of Saudi dialects. Aldarsoni (2011) surveyed the dialect distribution in Saudi Arabia. According to the documented variations, he found that there are five main dialects in Saudi Arabia (Najdi, southern, western, northern and eastern), and within each main dialect there are many sub-dialects. Aldarsoni’s taxonomy is based on the variations in lexis among Saudi dialects. Alrumaih (2002) examined dialect areas in Saudi Arabia using the perceptual dialectology method of the map-drawing task, in which participants were asked to draw the boundaries of regional varieties in Saudi Arabia. Alrumaih (2002) found the same dialect areas as in Aldarsoni’s survey. Similarly, Al-Rojaie (2021) used the map task but with more participants (674), and the study revealed the same dialect areas as in Aldarsoni and Alrumaih’s research (see Fig. 1).

Fig. 1: Saudi Arabic dialect map.
figure 1

Reproduced with permission of Al-Rojaie (2021).

Full size image

Thus, the following taxonomy of Saudi dialects was revealed in previous research (Al-Rojaie, 2021, p. 485), and will be followed in the current research (see Fig. 1):

  1. 1.

    Central region dialect (i.e., Najdi dialect), including cities such as Buraydah, Riyadh, Al Kharj, Hawtat Bani Tamim, Wadi Ad-Dwasir and Al Aflaj.

  2. 2.

    Northern region dialect, including cities such as Al Quryaat, Arar, Al-Jouf and Tabuk.

  3. 3.

    Eastern region dialect, including cities such as Hafar Albatin, Khafji, Al-Qatif, Jubail, Dammam and Hofuf.

  4. 4.

    Western region dialect (i.e., Hijazi dialect), including cities such as Yanbu, Jeddah, Taif and Makkah.

  5. 5.

    Southern region dialect, including cities such as Albaha, Bisha, Jazan, Najran and Abha.

As the aim of the present study is to explore the main dialects in Saudi Arabia, it will not address perceptions in relation to sub-dialects.

National picture

Saudi Arabia was established in 1932 by King Abdulaziz Al Saud and follows a monarchical system. Before the establishment of the Kingdom, the region was predominantly controlled by a feudal tribal system (Al-Rojaie, 2020b; Anishchenkova, 2020). Prior to the founding of Saudi Arabia, members of each tribe were connected by the same genealogy and were governed by tribal chiefs who had dominant political power over their tribes (Nahedh, 1989). People’s identities were characterised by their affiliations with their tribes, clans and regions. Thus, there was no national identity for Saudis before the establishment of the country; rather, the identity of each individual was strongly related to the region from which their tribe came (Al-Rojaie, 2020b).

After the unification of the Kingdom, the government attempted to support a national identity that had been missing in the Saudi social structure (Al-Rojaie, 2020b). In doing so, the government encouraged the national integration of the tribes by offering them many opportunities, including job opportunities, housing estates in the cities and educational opportunities for their children. With the discovery of oil in 1938, the country went through a period of extensive urbanisation and development (Albatel, 2005). Despite significant development in the country, the influence of the tribal system continues to exert a significant impact on the social fabric of Saudi Arabia, particularly in terms of regional segmentation based on tribal affiliations. In essence, alongside the presence of distinct regional identities among Saudis, tribal affiliations remain pervasive within Saudi society, thereby potentially shaping language attitudes towards various dialects.

The Najdi dialect is spoken in the Najd region in central Saudi Arabia. The name ‘Najd’ means ‘highlands’. Unlike other regions in Saudi Arabia, the Najd region is isolated from external contact as it is not open to external borders. Geographically, this isolated location means that this dialect is not as affected by language variation and change (Ingham, 1994). Thus, the Najdi dialect has been described as the closest to MSA (Anishchenkova, 2020) and is perceived, based on experimental evidence, as the most ‘correct’ dialect in Saudi Arabia (Alrumaih, 2002). Politically, Najd was the first region that the founder of Saudi Arabia occupied. Because the region witnessed the king conquering and unifying the whole Kingdom, Najd has a historical and political prominence over other regions in Saudi Arabia. After the establishment of the country, the King named the city of Riyadh, at the heart of Najd, the capital of the nation. From the establishment of the Kingdom through to the present day, this region has been the official home of the royal family; thus, the royal family speaks the Najdi dialect as its mother tongue (Al-Rojaie, 2021). Accordingly, the geographical location and political importance of Najd help the dialect be perceived as the most prestigious of the Arabian Peninsula (Aldosaree, 2016; Anishchenkova, 2020). In a recent study, Al-Rojaie (2020a) investigated the emergence of a new koine in Saudi Arabia in urban centres. He used a map-drawing task to reveal a final composite map that could visualise the dialect areas that are emerging as the new standard in Saudi Arabia. The findings revealed that the closest dialect to the koine is Najdi Arabic, ‘particularly the dialect of the capital city of Riyadh’ (Al-Rojaie, 2020b, p. 37).

The southern dialect is spoken in the southwest of Saudi Arabia. This region has a direct neighbour to the south, Yemen. Linguistic researchers have investigated the Yemeni sub-dialects of the southern region from a variationist approach and have revealed a notable resemblance between Yemeni and some southern sub-dialects (Watson, 2014). There are many territories in the southern region, such as Asir, Tihama and Jazan. Anishchenkova (2020, p. 13) describes Tihama as ‘underdeveloped in comparison to the rest of Saudi Arabia, and its population…among the poorest in the country’. Lowry (2021) investigated the language ideology in the Jazan territory and commented that ‘until the 1990s, it was one of the least developed regions in Saudi Arabia’ (p. 41). Habib (1988) explains why Jazan is viewed as the most underdeveloped region; he explains that low incomes, the late advent of higher education and weak infrastructure are among the factors that lead to the region being characterized as underdeveloped. Recently, after the launch of Vision 2030, the region has been developing quickly, as is evidenced by the many infrastructure and economic projects implemented in the region (Vision 2030 Projects, n.d.).

The northern dialect is spoken in the northern region, which is the least populated region in Saudi Arabia (Anishchenkova, 2020). The eastern region stretches along the coastal region near the Arabian Gulf. This region’s direct neighbours are certain Arabian Gulf countries, including Qatar, Bahrain, the UAE and Kuwait. The eastern dialect is sometimes referred to as the Gulf dialect because linguistically, this dialect is perceived to be a mixture of various Arabian Gulf dialects (Holes, 1990).

Stereotypical perceptions of Saudi dialects are driven by mass media. The representation of Saudi dialects points towards a structured and recurring pattern that downgrades some dialects in comparison to others. Saudi media depicts southern characters as naive and simple, Najdis as intelligent and shrewd and westerners as humorous and funny. Although this is anecdotal evidence, it points to the stereotypical image of Saudi dialects and how they are driven by media representations. The role of the media is pivotal and has been proven even to affect perceptions of linguistic variations in society. Montgomery (2012) investigated the effects of media on cultural prominence and how those effects raised participants’ awareness of dialectical variations in many distinct dialect areas in Great Britain. He found that cultural prominence, which is boosted by media representations of various dialects and accents in Great Britain, could be a key factor in raising the participants’ levels of dialectical variation, even if the dialects are not geographically adjacent to each other.

Methodology

Methods

The data for the current study were collected using a questionnaire, which was divided into two parts. The first part collected data pertaining to RQ1, and the second part elicited answers about the participants’ social backgrounds, which pertain to RQ2.

The study followed two widely used methods in the field. The first one is the VGT, which was used to present the voice samples. The technique has been widely used in the literature, which proves its effectiveness (see Huygens and Vaughan, 1983; Ladegaard, 1998). The second one is a semantic differential scale, which was used to rate each adjective according to ‘equidistant numbers on a scale (e.g. 1–7) with semantically opposing labels applied to each end’ (Garrett, 2010, p. 55). A semantic differential scale provides interval data, which are required in advanced, sophisticated analyses, such as FA. The technique has been widely used in the literature to reveal dimensional models in LAs (see Kristiansen, 2001; Zahn and Hopper, 1985).

Prior to designing the questionnaire, three participants from each of the four regions of Saudi Arabia (representing the four dialects in question) were selected for voice recordings. Their ages ranged from 35 to 45 years, and they were all men.Footnote 1 They were provided with a Saudi map showing the general weather in Saudi Arabia and were asked to explain the weather according to the provided map. Thus, instead of giving them a fixed written text to read from, they were asked to discuss the weather freely in their own dialects. The justification for using this technique is to allow the participants ‘to talk more spontaneously in their normal speech’ (Garrett, 2010, p. 63). The same technique was used by Huygens and Vaughan (1983), who allowed speakers to speak freely by giving them a map of directions and asking the speakers to describe the directions as drawn on the map.

It is important to note that with this technique it is difficult to control voice quality and reading and delivery style as the speakers were initially asked to speak freely. They were given instructions that helped exert some control over the variations between the recordings, such as speaking for exactly one minute. They were also asked to avoid background noise. After receiving the audio recordings, the researcher uploaded them into the Audacity softwareFootnote 2 to reduce background noise before uploading the samples to the questionnaire, which was designed using the QuestionPro online tool.Footnote 3

Each regional variety was represented and recorded by three speakers. Afterward, the researcher recruited a panel of judges from all regions (a total of eight) and asked them to select the best representative speaker of each region. The judges were chosen on the basis of two criteria: their length of residence and their affiliation with the regional tribes that represent each dialect area. They must have lived in the dialect area for at least 10 years, and they needed to be originally affiliated with the local tribe in each dialect area. Then, after the voice samples were chosen and ready, they were uploaded to the online questionnaire.

Although attempts were made to approach female speakers to obtain their consent to record their voices for use in the questionnaire, female speakers from the northern and southern regions refused to give their consent. They justified it by explaining that tradition would not allow them to do so as they came from conservative backgrounds. However, female speakers from the Najdi and eastern regions were willing to do so. Therefore, for reasons out of the researcher’s control, the samples of the speakers’ voices were confined to men only.

The first part of the questionnaire addressed RQ1. The speakers listened to each voice sample and were asked two questions. First, based on Preston (1999), participants were asked to assign the voice samples to their original areas. They were given five main regions in Saudi Arabia to choose from. In the second question, they were asked to rate the voice samples against 14 characteristics on a semantic differential scale for all four dialects, ranging from one to five (where one referred to the least and five to the most). The characteristics were adapted from preliminary research on LAs in Saudi Arabia by Alfalig and Alhazmi (2022). The characteristics that were adapted from their study were that the speaker sounded ‘Bedouin’, ‘prestigious’, ‘calm’, ‘beautiful’, ‘uncultured’, ‘harsh’, ‘arrogant’, ‘confident’, ‘friendly’, ‘open-minded’, ‘moralistic’, ‘easy to understand’, ‘arrogant’ or ‘difficult to understand’. The remaining characteristics in their study were discarded because they either recorded very few instances or were difficult to use with a semantic differential scale as they were not evaluative (e.g., terms that represented geographical locations or cultural symbols). Another reason for limiting the items in the questionnaire to only 14 was to avoid fatigue for the participants. Because participants were asked to listen to four guises and accordingly rate each guise in relation to 14 chosen items, in total they needed to give ratings for 56 items.

The second part of the questionnaire addressed RQ2. It elicited participants’ demographics, including their genders, ages, levels of education and places of birth. These formed the independent variables in the study. The reason for choosing these social demographics for inclusion in the study was that they have a long history of proven effectiveness in LA research (for age, see Paltridge and Giles, 1984; Yin and Li, 2021; for gender and location, see Montgomery, 2012; for educational level, see Kircher and Fox, 2019).

Procedures

The questionnaire was distributed online using Twitter and WhatsApp. Data collection took place in June and July 2021. The target population for the study was all Saudis living inside or outside Saudi Arabia. The sampling technique was convenience sampling. This is a type of non-probability sampling in which participants are chosen because they are easy to reach. One of the major disadvantages of this technique is that its results are less generalisable as it is difficult to reach a representative sample of the target population. However, in the current study, the convenience sample drawbacks have been compensated for by the use of an online questionnaire, which grants access to different demographic groups in many locations in Saudi Arabia, similar to the Saudi population.

Participants

The participant sample consisted of 568 Saudis. This sample is not representative of the population of Saudi Arabia, which is nearly 35 million, as the confidence interval, which is 95%, around the study mean overlaps with the true population mean. Furthermore, the margin of error in the current study is 4.15%.Footnote 4 The survey invitations totalled 2437 while actual responses totalled 568 (response rate of 23.31%).Footnote 5 Accordingly, given a confidence interval of 95%, a margin of error of 4.15% and a response rate of 23%, the required sample size should be 601.Footnote 6

Participants’ self-defined regional dialects, genders, ages, educational levels and places of birth are represented in Table 1.

Table 1 Summary of the subject pool.
Full size table

Statistical analysis

The data analysis comprised FA, reliability testing and ANOVA.

In stage one, FA was used to reveal the latent dimensions behind the associated characteristics. According to Conway and Huffcutt (2003, p. 150), ‘if a researcher’s purpose is to understand the [underlying] structure of a set of variables […], then use of a common factor model, such as principal axis or maximum likelihood factoring, represents a high-quality decision’.

FA goes through three phases before a final decision is made. The first phase involves running the analysis for the Kaiser–Meyer–Olkin (KMO) measure and Bartlett’s test. Both tests are conducted to ensure that the data is appropriate and suitable for FA (Pallant, 2013). The second phase involves extracting the factors, which relies on Kaiser’s criterion and a scree plot. The third phase involves rotating the factors. Factor rotation supports the final solution by combining similar variables together (Conway and Huffcutt, 2003). Furthermore, it prevents multiple loadings into the extracted factors (Thurstone, as cited in Pallant, 2013).

In stage two, a reliability test was conducted to ensure that the scales and dimensions had internal consistency (Pallant, 2013).

In stage three, an ANOVA test was used to reveal whether any of the independent variables (i.e., dialect, gender, age, education or place of birth) had an effect on the dependent variables (i.e., factor scores).

Results

The dimensional model findings

Allocation task results

Prior to rating each speaker, participants were asked to assign each speaker to their original place. Generally, the results revealed that most speakers were assigned correctly. The highest number of correct assignments were towards eastern speakers (91.9%), southern speakers (91.0%), northern speakers (87.7%) and Najdi speakers (83.3%) (see Table 2).

Table 2 Allocation results of the southern, northern, eastern and Najdi speakers.
Full size table

FA results before refining

The KMO measure of sampling adequacy and Bartlett’s test were used. Pallant (2013) suggests that the KMO test result should be 0.6 or above and that Bartlett’s test result must be significant for the data to be sufficiently suitable. In this study, the KMO test result was 0.806, which is acceptable, and Bartlett’s test sphericity value was significant (p = 0.000). Second, Kaiser’s criterion and a scree plot were employed. Kaiser’s criterion revealed three factors with eigenvalues exceeding one, and all explained 51.243% of the variances (F1 25.560, F2 13.150, F3 12. 534).Footnote 7 The scree plot revealed a sharp break after F2 (see Fig. 1). Both Kaiser’s criterion and the scree plot indicated a three-factor solution.

Reliability test

Cronbach’s alpha of the three factors recorded 0.712 for F1, 0.638 for F2 and 0.661 for F3. It is suggested that if Cronbach’s alpha value is <0.7, the researcher is recommended to delete any item that recorded a value that is higher than the final alpha value if the item was deleted (Pallant, 2013). The items ‘Bedouin’ in F2 and ‘arrogant’ in F3 were 0.707 and 0.691, respectively, if they were deleted from the scale. Therefore, they were deleted from the F2 and F3 scales to obtain higher values. Consequently, after the deletion of the two items, the F2 value went up to 0.707, and the F3 value went up to 0.691.

Thus, the FA needed to be re-run as ‘Bedouin’ was deleted from F2, and ‘arrogant’ was deleted from F3.

FA results after refining

The KMO test result was 0.812, which is acceptable, and Bartlett’s test sphericity value was significant (p = 0.000). Second, using Kaiser’s criterion and a scree plot, Kaiser’s criterion revealed three factors with eigenvalues exceeding one, and they explained 52.798% of the variances (F1 29.220, F2 13.533, F3 10.045). The scree plot revealed a break after F3 (see Fig. 1). Both Kaiser’s criterion and the scree plot indicated a three-factor solution (Fig. 2).

Fig. 2
figure 2

Scree plot test.

Full size image

Factor rotation and interpretation

As the factor correlation matrix indicated that the extracted factors were correlated, the oblique rotation (direct oblimin) rotated solution was acceptable (Pallant, 2013) (see Table 3).

Table 3 Factor correlation matrix.
Full size table

After factor rotation, the three-factor solution was ready to be introduced (see Table 4). Thus, five evaluative terms were loaded in F1 with a mean score of 3.30, four were loaded in F2 with a mean score of 3.62 and four were loaded in F3 with a mean score of 3.47. F1 and F2 seemed to share many similarities with the dynamism factor in the LA literature. Therefore, both F1 and F2 were treated as two-type dynamism factors with some minor differences between them. More specifically, the characteristic that shared the most variance with F1 was ‘prestigious’, which finds its way into the status dimension. Previous research has revealed that ‘prestigious’ is strongly indexical of the status dimension (see Garrett, 2010; Garrett et al., 2003; Kircher and Zipp, 2022); thus, F1 is named dynamism-status to encompass more hints of the status-type scale. This is because the current study reveals an overlap between the characteristics of dynamism and status within one dimension; therefore, it is named dynamism-status. The same applies to F2, which is named dynamism-language because the factor shares some characteristics with the dynamism dimension, and two main characteristics are language-centred: ‘the dialect is easy to understand’ and ‘the dialect is difficult to understand’. This dimension encompasses characteristics related to the attitudes towards dialect comprehensibility, along with attitudes towards dialect speakers.

Table 4 Rotated factor matrix.
Full size table

F3 seemed to share many similarities with the solidarity factor revealed in LA research (see Garrett, 2010; Garrett et al., 2003; Kircher and Zipp, 2022), so it is named solidarity.

Dialect means scores on the three factors: association of dialect with the dimensions

The study separated the data for each dialect and calculated the mean scores within each factor to reveal how each dialect was perceived in accordance with the new factors. The FA identified three dimensions. Univariate statistical analysis revealed that the four dialects in the study were judged differently from each other in the three dimensions (see Table 5).

Table 5 Means and standard deviations of Saudi participants’ judgments of Saudi dialects (1 = least, 5 = most).
Full size table

As shown in Table 5, the Najdi dialect was seen as having the fullest evaluative profile in the two-type dynamism factor, with mean scores of 3.70 and 3.86, respectively, followed by the northern (3.44, 3.74), eastern (3.35, 3.64) and southern, which was seen as lacking dynamism (2.69, 3.22). Regarding F2, participants’ judgements revealed a very similar pattern to the dynamism-status factor. As for the solidarity factor, the northern dialect was viewed as carrying the most solidarity (3.59), while the eastern dialect was seen as having the least solidarity (3.32).

Overall, each dialect proved to have a different evaluative profile.

Demographic effects findings

Dialect

In the dynamism-status dimension, the southern and Najdi dialects recorded significant results (see Table 6). There was a statistically significant difference at the p < 0.05 level for the southern and Nadji dialects. Post-hoc comparisons using the Tukey HSD test indicated that the participants whose original dialects were southern and Najdi gave their own dialects higher mean scores in dynamism status compared to those whose dialects were northern, eastern, Najdi or Hijazi (see Table 7).

Table 6 Effects of the participants’ own dialect on the results.
Full size table
Table 7 Post-hoc test results on the dialect demographic variable.
Full size table

In the dynamism-language dimension, the Najdi dialect recorded significant results (see Table 6). There was a statistically significant difference at the p < 0.05 level for the Najdi dialect. Post-hoc comparisons using the Tukey HSD test indicated that the participants whose original dialects were Najdi and Western gave the Najdi dialect higher mean scores in dynamism language compared to those whose dialects were northern, southern or eastern (see Table 7).

A different pattern that emerged in the dynamism-language dimension was that eastern and northern dialects recorded significant results (see Table 6). There was a statistically significant difference at the p < 0.05 level for the southern dialect. Post-hoc comparisons using the Tukey HSD test indicated that the participants whose original dialects were Najdi and western gave the northern and eastern dialects higher mean scores in dynamism language compared to those whose dialects were northern, southern or eastern (see Table 7). This is regarded as a different pattern because the speakers of the northern and eastern dialects did not rate their dialects highly, as opposed to the Nadji participants who confidently rated their dialect high in the same dimension.

Gender

Regarding dynamism status, the male mean score was higher than the female mean score towards the perception of the southern dialect. The pattern is reversed when it comes to the Najdi dialect: the female mean score was higher than the male mean score regarding the perception of the Najdi dialect at p < 0.001 (see Table 8).

Table 8 Effects of the participants’ gender on the results.
Full size table

Regarding dynamism-language, the female mean score was higher than the male mean score regarding the perception of the southern, northern, eastern and Najdi dialects (see Table 8).

Regarding solidarity, the female mean score was higher than the male mean score regarding the perception of the southern, northern, eastern and Najdi dialects (see Table 8).

Level of education

a significance level of p < 0.05 was found among secondary, undergraduate and postgraduate education levels regarding the perception of southern, northern, eastern and Najdi dialects (see Table 9).

Table 9 Effects of the participants’ level of education on the results.
Full size table

For dynamism-status, post-hoc comparisons using the Tukey HSD test indicated that Najdi participants whose level of education was undergraduate gave their dialect higher mean scores in the dynamism-status dimension than students at the secondary level of education (see Table 10). Conversely, southern participants whose level of education was secondary gave their dialects higher mean scores in the dynamism-status dimension than postgraduates (see Table 10).

Table 10 Post-hoc test results on the level of education demographic variable.
Full size table

For dynamism-language, post-hoc comparisons using the Tukey HSD test indicated that postgraduate participants gave the eastern, northern and Najdi dialects higher mean scores compared to those whose level of education was secondary.

In the solidarity dimension, post-hoc comparisons using the Tukey HSD test indicated that southern participants whose level of education was undergraduate gave their dialects higher mean scores than those who only had secondary education (see Table 10).

Place of birth

Regarding the dynamism-status dimension, a significance level of p < 0.001 was found regarding the perception of the southern and Najdi dialects (see Table 11). Post-hoc comparisons using the Tukey HSD test indicated that the participants whose birthplaces were in the southern and Najdi regions of Saudi Arabia gave their own dialects higher mean scores compared to those whose birthplaces were in the eastern, northern and Hijazi regions of Saudi Arabia (see Table 12).

Table 11 Effects of the participants’ place of birth on the results.
Full size table
Table 12 Post-hoc test results on the place of birth demographic variable.
Full size table

Regarding dynamism-language, a significance level of p < 0.001 was found regarding the perception of the northern, eastern and Najdi dialects (see Table 11). First, post-hoc comparisons using the Tukey HSD test indicated that the participants whose birthplaces were in the eastern, Najdi and western regions of Saudi Arabia gave the northern dialect higher mean scores compared to those whose birthplaces were in the southern and northern regions of Saudi Arabia. Second, post-hoc comparisons using the Tukey HSD test indicated that participants whose birthplaces were in the eastern, Najdi and western regions of Saudi Arabia gave the eastern dialect higher mean scores compared to those whose birthplaces were in the southern and northern regions of Saudi Arabia. Third, participants whose birthplaces were in the Najdi region of Saudi Arabia gave the Najdi dialect higher mean scores compared to those whose birthplaces were in the northern, southern and eastern regions (see Table 12).

Regarding the solidarity dimension, a significance level of p < 0.001 was found regarding the perception of the southern dialect (see Table 11). Post-hoc comparisons using the Tukey HSD test indicated that participants whose birthplaces were in the southern and western regions of Saudi Arabia gave the southern dialect higher mean scores compared to those whose birthplaces were in the northern regions of Saudi Arabia (see Table 12).

Discussion

The dimensional model: implications for LA theory

The three-dimensional model

The present study reveals a three-dimensional model that is mainly driven by the dynamism dimension. This is because two of the dimensions find their way into the dynamism dimension. All in all, the study found a two-type dynamism dimension: dynamism-status and dynamism-language. The third dimension revealed in the model is solidarity.

Previous research over several decades and in many parts of the world has revealed a consensus that LAs are bidimensional and usually operate between status and solidarity (El-Dash and Busnardo, 2001; Genesee and Holobow, 1989). Recently, the dynamism dimension has been revealed as a separate dimension to be added to status and solidarity (Grondelaers et al., 2019; Kristiansen, 2001, 2009; Zahn and Hopper, 1985). The current research revealed a three-dimensional model with dynamism as the predominant dimension instead of status.

The pattern suggests that the status-solidarity evaluative distinction is not significant to Saudis’ evaluations of their dialects; rather, dynamism-solidarity seems to be present in their minds. Previous studies reveal that status versus solidarity is exposed when there is a standard and non-standard language being evaluated (Kristiansen, 2001). In Saudi Arabia, Najdi was perceived as the koine/standard dialect in a recent perceptual study (Al-Rojaie, 2020b); thus, based on experimental evidence, Najdi is the new standard language in Saudi Arabia. However, with the emergence of a new standard in Saudi Arabia, the solidarity-versus-status pattern is not revealed in the current study. Two interpretations can be put forward. It might be that the dialect has gained its standardness recently, in which case dynamism seems to emerge strongly with new standards, as is the case with the emergence of neo-standards in Copenhagen, where similar results obtained indicate that dynamism versus solidarity is more important than status versus solidarity (Kristiansen, 2009). The second interpretation is related to the rapid urbanisation Saudi Arabia has been going through recently, embracing social and cultural reforms. Thus, the dynamism of a language can be seen as indexical of social and cultural reforms occurring in society. Hence, with the spread and speed of urbanisation and development all over the world, and in the Middle East in particular, the study posits that the dynamism dimension might be the driving force in LA findings, rather than the status dimension, in fast-growing societies that are experiencing transformational shifts from conservative to open trends, subject to verification in the future.

It is possible that the strength of the dynamism dimension revealed in this study is due to the large number of characteristics that were initially inserted in the questionnaire that refer to the dynamism and vitality of the dialects. According to the findings, this is unlikely because the characteristics inserted in the questionnaire were taken from a previous study that aimed to elicit broad LAs in Saudi Arabia. Further evidence comes from Al-Rojaie (2021), as mentioned in section 4. The study revealed an extensive number of dialect attitudinal characteristics in Saudi Arabia, most of them reflecting dynamism and social attractiveness.

The semantics of the model

The dynamism-status dimension includes characteristics such as ‘open-minded’, ‘confident’, ‘prestigious’, ‘easy to understand’ and ‘beautiful’. The characteristics of ‘prestigious’ and ‘beautiful’ seem to be related to different attitudinal dimensions as revealed in the literature. First, the characteristic ‘prestigious’ has widely been associated with the status dimension (see El-Dash and Busnardo, 2001; Genesee and Holobow, 1989). It might be that from the point of view of the Saudis, one of the indexical qualities of a dynamic dialect is the dialect having some status and being prestigious. Second, with regard to the characteristic ‘beautiful’, it is important to mention that this characteristic is indexical of aesthetic quality, which has been widely associated with the solidarity dimension in previous studies (see Yin and Li, 2021; Zahn and Hopper, 1985). However, according to Saudis’ ideological suppositions, the beauty of a dialect is associated with the dynamism dimension.

The dynamism-language dimension includes characteristics such as the dialect being difficult to understand, easy to understand, cultured or harsh. This dimension is named dynamism-language as the top two characteristics that shared the most variance in this dimension are attitudes towards the perceived ease or difficulty of a dialect. Previous research in LAs has revealed a strong connection between language use and dynamism (Rezaei et al., 2017; Yin and Li, 2021). What follows is that the comprehensibility of the language or dialect is a basic factor for the language user to use it. Therefore, the comprehensibility of the dialects in this study is viewed as a trigger factor for their dynamism, along with the other characteristics in the dimension. One might think that three of the characteristics (i.e., difficult to understand, harsh and uncultured) are presented in a negative way in this dimension, but this is unlikely to be the case as the scale is reversed in statistics to prevent response bias. Thus, high scores in both dynamism dimension types indicate high dynamism and low scores indicate low dynamism, regardless of whether the characteristic is negatively worded.

The solidarity dimension includes ‘generous’, ‘moralistic’, ‘calm’ and ‘friendly’. These characteristics are in line with characteristics revealed in previous research (Garrett, 2010).

The association of Saudi dialects with the dimensional model

Two main findings emerged here; the Najdi dialect is highly associated with the two-type dynamism dimension, recording the highest mean scores in both dynamism dimensions, whereas the southern dialect recorded the lowest mean scores in both dynamism dimensions. This result indicates that the Najdi dialect is perceived as the language of dynamism in Saudi Arabia. This result can be interpreted by two justifications. First, the Saudi Vision 2030 is seen as a new dynamic construct that launched in Riyadh, which is the capital of Saudi Arabia and located at the heart of the Najdi region. Therefore, contextually, as the call for economic, cultural and social reforms originated in the Najdi region, its people were the first to accept the changes. Accordingly, their dialect is perceived as the most dynamic.

Second, the mean scores of the associations of the Saudi dialects in the solidarity factor show that the mean difference is slight, ranging between 3.32 and 3.59. Therefore, it seems that although the society experiences dynamic changes at all levels, a sense of solidarity is maintained in Saudis’ language ideology. It is worth mentioning here that the northern dialect has the fullest evaluative profile in solidarity, which might be due to the fact that this region specifically is the least populated. This could potentially positively affect its people’s solidarity.

Effects of demographics: implications for media and language planning

Native dialect, gender, level of education and place of birth significantly affected the results. First, southerners and Najdi rated their dialects higher than any other participant did for their dialect in the dynamism-status dimension.

Second, female participants rated the northern, eastern and Najdi dialects higher than male participants in the dynamism-status dimension. Furthermore, women rated southern, northern, eastern and Najdi dialects higher than the male participants in the dynamism-language dimension and solidarity dimension.

Third, participants with secondary educations rated the southern dialect higher in the dynamism-status dimension than postgraduates. Conversely, participants with undergraduate educations rated the Najdi dialect higher in the dynamism-status dimension than participants with secondary educations. Furthermore, undergraduates and postgraduates rated the northern, eastern and Najdi dialects higher than other participants with secondary educations in the dynamism-language dimension. Finally, participants with undergraduate educations rated the southern dialect higher in the solidarity dimension than participants with secondary educations.

Fourth, southerners whose places of birth were in the south region rated their dialect higher in the dynamism-status dimension and solidarity dimension, and Najdis did the same for their dialect in the dynamism-language dimension and dynamism-status dimension. It is important to note here that the northern and eastern dialects also received higher ratings in the dynamism-language dimension, but the high ratings came from participants with diverse places of birth: eastern, Najdi and Hijazi regions. In other words, the higher ratings for southern and Najdi dialects came only from those who were born in these places, unlike the pattern revealed for eastern and northern dialects.

These four main findings will be discussed in more detail. The first and fourth findings will be discussed together as they reveal a similar pattern of in-group loyalty. The second and third patterns will be discussed separately.

The effects of native dialect and place of birth demonstrate that the southern and Najdi participants showed in-group loyalty. This raises the question of why the competitive pattern is only revealed between these two dialects. Two interpretations can be put forward here. Regarding the southern dialect, it is argued that the role of the media has played a crucial part in forming this ideology in Saudis’ wider beliefs. The Saudi media has depicted southerners as naïve and simple (see section “National picture”). It might be that southerners are attempting to modify such negative stereotypes by favouring their own dialects in both dimensions. It is suggested that the Saudi media needs to reconsider the representation of social images based on the real practices of social groups in society, rather than on presuppositions. On the other hand, Najdis could be reinforcing their prominence among other social groups in Saudi Arabia. The association of in-group loyalty to the two-type dynamism dimensions that emerged in the current results contradicts previous research, in which in-group loyalty has always been associated with the solidarity dimension (Dragojevic et al., 2013). In the present research, it is evident that the pattern is strongly associated with the two-type dynamism dimension.

Regarding the effect of gender, it was evident that female participants were more positive towards the Najdi dialect in the dynamism-status dimension than male participants. Furthermore, female participants were also more positive towards southern, northern, eastern and Najdi dialects in the dynamism-language and solidarity dimensions than male participants. This finding suggests that female participants are more accepting and accommodating towards dialect variations compared to men. In a similar vein, in multilingual research, many previous studies have demonstrated that female participants show more positive attitudes towards multilingualism (see Lai, 2007; Wright, 1999). In LA studies, this result echoes Bishop et al.’s (2005) study on attitudes towards different English accents, in which they found that women accorded higher positive attitudes than men to the prestige and social attractiveness dimensions.

The final pattern that emerged from the results shows a dichotomy between education levels. The pattern towards the southern dialect suggests that first, the less educated the participants were, the higher ratings they gave to dynamism- status dimension. Second, The pattern towards the Najdi, eastern and northern dialects implies that the more educated the participant was, the more positive attitudes they gave to both dynamism status and dynamism language dimensions. Third, the more educated the participants were, the more positive attitudes they attributed to the southern dialect in the solidarity dimension. Thus, the findings suggest that more positive attitudes are increased by increased levels of education. Similarly, Kircher and Fox (2019) found that participants with high levels of education delivered positive attitudes to multicultural London English (MLE) dialects in London in the item ‘for MLE speakers, using MLE is an important part of being a young Londoner’ (Kircher and Fox, 2019, p. 855).

Thus, the study suggests that an education-related pattern can be generalised from the results. This pattern postulates that an educated person has less biased attitudes towards Saudi dialects. In this regard, the study recommends working on an educational programme that aims to educate students about the value of linguistic and cultural diversity. In other words, the programme teaches students how to appreciate and respect the various dialects spoken in the country. The programme should be launched from the educational system in the country as one of the results regarding the effects of socio-demographics indicates that an increased level of education would decrease participants’ bias regarding their dialects. Therefore, the current study argues that if the role of Saudi dialects in enhancing the country’s culture were to be implemented in education, it could reduce prejudiced LAs in the country.

Limitations

The findings are subject to one key limitation: the speakers were mainly men. Although every effort was made to obtain a proportional number of men and women, this was unsuccessful. Specifically, all the female speakers of the northern and southern dialects refused to have their voices recorded and distributed online. When first approached, they willingly desired to take part in the study, but they withdrew after being informed that their voices would be incorporated into an online questionnaire and distributed online. The study suggests that could female voices be recorded, they would evoke more pronounced attitudes, especially in the dynamism dimension. Grondelaers et al. (2019) found that the gender balance in their attitude experiment contributed greatly to providing a vivid picture of the dynamism dimension as female speakers are ‘known to spearhead nonstandard innovations, and [such] innovations are typically deemed inferior but dynamic in the first stage of their emergence’ (Grondelaers et al., 2019, p. 232). Therefore, further gender-balanced LA research is needed in the Saudi context to provide new insights into the discussion of dynamism.

Second, the sample size is not representative. Although the questionnaire was distributed online, the sample size did not reach the representation cut-off point. Given the above two limitations, the findings of the study cannot be generalised.

Third, the Hijazi dialect has been excluded from the study, which might further affect the generalisability of the study. It has been excluded because the linguistic situation in Hijaz is complicated when it comes to defining the Hijazi (western) dialect. This is because the region has a unique dichotomous dialect situation that operates between Bedouin Hijazi Arabic and Hadari Hijazi Arabic due to many religious and economic reasons that have hugely affected the migration to that region (see Alhazmi, 2018). Saudi media always depicts the Hadari Hijazi dialect as the representative dialect of Hijaz and neglects the presence of the other main dialect in Hijaz, Bedouin Hijazi. Experimental evidence has confirmed that Hadari Hijazi is viewed as the typical dialect of Hijaz, neglecting the Bedouin dialect (see Alahmadi, 2016; Alhazmi, 2018). Therefore, if the dialect were to be included in the study, there must be two speakers representing the Hijazi dialect, one from Bedouin Hijazi and one from Hadari Hijazi. As a result, overlapping in the dialect assignment findings would be expected as I introduced only one representative speaker from each region in Saudi Arabia. When it comes to Hijaz, the respondents would encounter a Hadari speaker that they would easily recognise and assign to the western region, but Bedouin Hijazi might present a problem as the dialect is not always represented in the media and might not be recognised by all the participants. Therefore, I preferred to avoid the Hijazi dialect as the linguistic situation is too complex, and it is better to investigate it with much more focus on the region itself.

The linguistic situations of the other salient dialects in Saudi Arabia are completely different from the ones in Hijaz. Take the Najdi dialect as an example; the dialect has sub-dialects in Najd, but all of them linguistically belong to each other as they share similarities in their phonological, phonetic and lexical features. By contrast, the linguistic situation in Hijaz is unusual.

Conclusion

This study revealed a three-dimensional model from an unexplored context. The model makes a promising contribution to LA theory. It reveals that dynamism instead of status is a powerful dimension in fast-growing countries such as Saudi Arabia. Furthermore, the association of the Saudi dialects revealed a dichotomous pattern between the Najdi and southern dialects in the dynamism dimension. It indicates that LAs seem to play a role in structuring Saudi society by operating in two dimensions: dynamism and solidarity. The dimensional model was further tested against the effects of social factors, and it was revealed that they highly affected the dynamism dimension, while the solidarity dimension revealed a more homogeneous pattern. The effects of social factors imply that the role of social media in Saudi Arabia has to be reconsidered to represent the social image of Saudi dialects based on their real practices rather than presupposed stereotypes. Furthermore, the findings suggest that language planners need to incorporate teaching about how diversity in Saudi dialects enhances the country’s culture and identity.