Children’s Multilectal Repertoires: Diglossic Style-Shifting by Palestinian Children and Adolescents in Syria

Abstract

Arabic diglossia, whereby Standard Arabic (SA) exists alongside numerous vernaculars, often leads to diglossic style-shifting, based on context or topic changes and marked in the vernacular by shifting to standard linguistic features. While this phenomenon has been widely studied in the speech of educated adults, research on diglossic style-shifting by children and adolescents has been rather limited. This paper investigates how it operates amongst 3–17-year-olds from a Bedouin speech community of Palestinian refugees in Syria. It examines context effects on realizations of the variables (θ) and (ð), which overlap with local realizations and (q), which has a standard realization ([q]) that is independent of dialectal variation in the community. Participants were recorded during sociolinguistic interviews and a picture-naming task, the latter being expected to evoke a school setting and prompt the use of more standard realizations, signaling diglossic style-shifting in their speech. Style-shifting was influenced by age, context, and the linguistic variables under examination. While picture-naming prompted greater use of standard realizations of all variables, shifting to [q] also appeared during the interview in lexical borrowings from SA, revealing topic effects on diglossic style-shifting. Children aged 6–14 exhibited more style-shifting in picture-naming, likely reflecting the central role of school in their lives, while the speech of 15–17-year-olds contained more lexical borrowing with [q]. This likely reflects their larger linguistic repertoires and longer exposure to SA than their younger peers. These findings indicate that SA plays a key role in participants’ linguistic practices and reflect their awareness of how to employ it appropriately in their speech.

1. Introduction

Sociolinguistic research has paid considerable attention to stylistic or intraspeaker variation, i.e., linguistic variation at the level of individual speakers in response to situational factors.

Early work in this area was pioneered by William Labov (1966) and framed as attention to speech. Labov (1972, p. 112), for example, asserts that individuals are not monostylistic, suggesting that speech styles can be arranged on a continuum based on the speaker’s level of attention.

Styles are, thus, subject to formality and speakers are presumed to use more standard-like and prestigious features when they are conscious of their speech, reflecting community-level interspeaker variation (Hernández-Campoy and Cutillas-Espinosa 2012). Despite being foundational to sociolinguistic theory and instrumental in the design of its methodology (e.g., Labov 1972; Trudgill 1972), attention-based style-shifting has received considerable criticism as being unidimensional and lacking explanatory power (Hernández-Campoy and Cutillas-Espinosa 2012; Kiesling 2009; Sharma 2018; Schilling-Estes 2008). Other models were proposed, with a focus on interpersonal relationships between speakers (e.g., Giles’ 1973 ‘accent mobility’). This focus was further expanded in Bell’s (1984, 2001) audience design model. Bell also points to topic effects on stylistic variation, augmenting Labov’s unidimensional model of style-shifting (Schilling-Estes 2008).

Labov’s (1972) ‘vernacular principle’, which underlies his paradigm of style-shifting and posits that the least conscious form of speech is the only ‘genuine’ vernacular, has also been criticized (Schilling-Estes 2008, p. 974). Instead, speakers are understood to possess and use a wide range of styles in various situations and for a variety of purposes regardless of attention to speech (Eckert 2000; Milroy and Gordon 2008, pp. 49–51; Schilling-Estes 2004). As such, in later accounts of stylistic variation, especially third-wave variationist research, style is understood as a resource and a practice rather than a mere reaction to external factors (e.g., Drummond 2018; Eckert 2012; Snell 2018). These accounts are ‘speaker centered’ rather than ‘style focused’ (Kiesling 2009, p. 171) and depict stylistic variation as a ‘proactive’ rather than ‘responsive’ practice through which speakers engage in the strategic use of language to construct and index different identities and social meanings (Coupland 2007; Eckert 2012; Hernández-Campoy and Cutillas-Espinosa 2012, p. 4). For example, according to Schilling-Estes (1998, p. 69), while shifts in speakers’ styles may be triggered by contextual changes, more often than not, they themselves ‘serve to bring about contextual changes.’ With this view, style can be performed through any number of linguistic variables that indicate a social identity on the individual level or beyond (Eckert 2012; Eckert and Rickford 2001). The choice of these variables is based on the indexical meanings assigned to them and which are assumed to be shared by speakers and hearers (Eckert 2004, pp. 42–44). They are fluid and flexible and subject to constant change, making style also flexible rather than static and giving the same variables different associations in different social constructs (Eckert 2004, p. 43). As such, speakers’ use of sociolinguistic variables in style variation indicates a conscious, intentional attempt to employ or indicate identity marking.

However, as Sharma (2018, p. 2) points out, while these ‘performative’ approaches to stylistic variation have contributed to a more nuanced understanding of style, they ‘risk implicitly assuming equal control of variants in an individual’s repertoire.’ Indeed, speakers’ engagement in stylistic variation is subject to linguistic competence, since knowledge of multiple styles or speech patterns does not necessarily translate into the ability to use them; as such, knowledge can be active for some speakers but passive for others (Andersen 1992). This competence can be governed by age, gender, and level of education or other external factors (Andersen 1992). On the other hand, speakers may style-shift even without having full command of the intended style.

Notwithstanding the warranted criticism of the original conceptualization of stylistic variation and the insightful and informative reformulations of style, context-based style-shifting remains a productive and useful paradigm that has yielded consistent results. It is also readily applicable to different language settings, especially where the lines between standard and colloquial are much more sharply defined, as in the Arabic context (e.g., Ferguson 1959, 1991; Mejdell 2006).

2. Style in the Context of Arabic

Arabic presents a classic case of diglossia, whereby a ‘genetically related’ but highly distinct standard exists alongside the spoken vernaculars (Mejdell 2006, p. 1). Despite being related and sharing many features that span all aspects of the language, such as root-based morphology, Standard and Vernacular Arabic diverge sharply on all levels of the language such as lexis, morphology, phonetics/phonology, and syntax (Abu-Rabia 2000; Albirini 2016; Ibrahim 2009; Saiegh-Haddad 2003). It is generally accepted that Standard Arabic (SA) is not natively spoken or acquired naturally by any speakers (Abu-Rabia 2000; Mejdell 2006). First and foremost, it is a written language with a tightly prescribed grammar and a highly codified orthography and is the official language in all Arabic-speaking countries (Albirini 2016; Miller and Caubet 2009). Vernacular Arabic (VA), on the other hand, is used as the primary spoken medium and remains largely unwritten and with no codified orthography (Albirini 2016; Haeri 2000; Ibrahim 2009; Mejdell 2006). The spoken vernaculars are natively acquired by children, whereas official and consistent exposure to SA comes later through formal education at school age (Abu-Rabia 2000; Saiegh-Haddad 2003).

Much like style, diglossia in Arabic has gone through many reformulations over the years. The original conceptualization of diglossia in Ferguson’s (1959) seminal work is largely contextual and posits a sharp divide between the standard written variety and the plethora of spoken vernaculars. While the two forms of language exist side by side, they are in complementary distribution such that SA is reserved for formal contexts including education and administration while the spoken vernaculars are restricted to informal and more intimate settings. This sharp dichotomy is reconciled by Ferguson’s (1991) acknowledgment of the many intermediate forms ranging from most to least standard that exist between either pole of the diglossic continuum, and his reconceptualization of diglossia, which places more emphasis on register variation between SA and the spoken vernaculars (Mejdell 2006).

In addition to being criticized as being ‘crude’ and ‘oversimplified’ (e.g., Mejdell 2006, p. 2), the contextual delineation of diglossia in Ferguson’s original model has also been challenged, and more emphasis on a functional distinction has emerged (e.g., Albirini 2011; Mejdell 2006). In this model, both SA and VA can occur in the same context to fulfil different functions of ‘varying levels of importance, complexity, and seriousness’ (Albirini 2016, p. 20). It is important to note that Ferguson’s original model, though largely contextual in its classification of SA and VA, also includes function. Both context and function, however, perfectly overlap in his model, creating a rigid divide between SA and VA with little room for overlap.

The coexistence of SA and VA in the Arabic context has led to speakers alternating and mixing between the two forms in diverse settings and for various purposes. This has received considerable attention in Arabic sociolinguistic research and produced a large body of literature. Many of the earlier studies, which emerged in response to Ferguson’s (1959) rigid dichotomous model, approached this phenomenon by proposing various intermediate forms that are not bound by a strict contextual divide (e.g., Badawi 1973; Mitchell 1986; Youssi 1995). Among these, Educated Spoken Arabic (ESA) (Mitchell 1986) features most heavily in the literature and is used to refer to the form of language used by educated speakers in formal settings. Later studies examined this phenomenon under the framework of diglossic codeswitching (CS) (e.g., Alaiyed 2018; Albirini 2011; Bassiouney 2006). Most research on diglossic CS has focused on what Albirini (2016, p. 240) refers to as ‘monitored speech’, which is likely to trigger the use of SA, such as public speeches, lectures, political interviews, religious sermons, and panel discussions (e.g., Alaiyed 2018; Bassiouney 2006; Mejdell 1999; Soliman 2008). The alterations between SA and VA in these contexts, which are reserved for SA in Ferguson’s (1959) model, are largely functional, whereby VA may be used to clarify certain ideas or signal a shift in seriousness. Most studies on diglossic CS support this functional divide between SA and VA (e.g., Albirini 2011; Bassiouney 2006; Chakrani 2015; Holes 1993; Mejdell 2006), which reflects their statuses as High and Low varieties (Albirini 2016). Moreover, as pointed out by Albirini (2016, p. 61), the Standard variety is never fully utilized in spoken discourse, as the command of SA fluctuates considerably between speakers (Mejdell 2006).

Under the framework of CS, diglossic codeswitching is referred to as ‘SA-VA codeswitching’ (Albirini 2016, pp. 224–26), with the assumption that educated Arabic speakers can sustain SA in conversation. However, we take the view that SA is not on the same continuum as spoken vernaculars but rather an important stylistic resource for Arabic speakers. As such, we adopt the term style-shifting in this paper (see Ervin-Tripp 2001; Labov 1966; Lahlou 1991). As noted above, situational style-shifting lends itself rather well to the context of Arabic, whether we consider the contextual or the functional conceptualizations of diglossia. Insights from social constructionist approaches to style are also useful in examining diglossic style-shifting, since speakers may use SA and VA to perform different identities or project various attitudes (e.g., Albirini 2011; Bassiouney 2006; Holes 1993).

Synthesizing these approaches to style and drawing on the methodology of the original Labovian paradigm, whereby different speech styles are elicited using specific tasks in particular situations, we investigate diglossic style-shifting in the speech of Palestinian children and adolescents in Syria.

3. Style-Shifting in the Speech of Children and Adolescents

3.1. Age

As stated above, most studies on diglossic style-shifting are carried out on monitored speech such as public speeches, lectures, and religious sermons. As such, diglossic style-shifting has, thus far, largely been examined in the speech of adult educated Arabic speakers, since competence in SA is a key factor in such switches (Albirini 2016; Hudson 2002; Mejdell 2006). Studies on diglossia involving children focus instead on topics related to literacy and reading acquisition (e.g., Abu-Rabia 2000; Saiegh-Haddad 2003, 2012) and little research has ensued on the stylistic repertoires of Arabic-speaking children and adolescents. This study, therefore, aims to examine the role of SA in Arabic-speaking children’s and adolescents’ linguistic behavior and development. By doing so, this research will enhance our understanding of the sociolinguistic competence of much younger speakers. Moreover, further empirical work is required on the sociolinguistic implications of diglossia on Arabic-speaking communities and the role of SA in language variation and change, considering the mostly ‘…impressionistic evaluations, which have characterized much of the discussion surrounding this topic’, as observed by Albirini (2016, p. 62).

Although early accounts of child language assumed late acquisition of social constraints on speech (e.g., Labov 1964; Lakoff 1973), children’s ability for stylistic variation has, in fact, been shown to emerge quite early in the acquisition process. Indeed, stylistic constraints on speech are acquired alongside, and in some cases prior to, grammatical constraints, as they form an integral part of children’s communicative needs (e.g., Chambers 2009; Foulkes et al. 2001; Johnson and White 2020). Research suggests that children as young as two or three years of age can vary their styles depending on perceived contexts (e.g., Kaiser 2022; Leaper 1991; Paugh 2005; Youssef 1993). Paugh (2005) examined patterns of codeswitching by 2–4-year-olds who were unbalanced Patwa–English bilinguals and concluded that children were able to demonstrate appropriate stylistic use of both languages in their role play, indicating an awareness of the role and association relevant to each of them. African American children were also found to decrease the use of vernacular features in their speech when they went to school (Van Hofwegen and Wolfram 2010), which implies an awareness of the association between Standard English and such settings.

Several early studies also show evidence of interesting patterns of social, stylistic, and linguistic variation in the speech of children (e.g., Fischer 1958; Purcell 1984; Romaine 1978). For example, Fischer (1958) found social variation in the alternation of [in] and [ing] in the production of 3–10-year-old children, whereby girls were found to use the standard [ing] variant more than boys. He also noted stylistic variation in the speech of a ten-year-old boy who overwhelmingly used [ing] in the formal interview but mostly used vernacular [in] in the informal one. Likewise, Purcell (1984) found social and stylistic variation in the speech of 5–12-year-old Hawaiian children, who switched between Hawaiian and General American English in response to factors such as a change in topic or addressee. Similarly, Romaine (1978) observed age, gender, and stylistic variation in the production of word-final –r among 6–8- and 10-year-old children who spoke a variety of Scots, with girls favoring a prestige variant and boys preferring a covertly prestigious one. Although results from these studies provide early evidence of children’s awareness of social and stylistic factors, they need to be approached with caution, as some of the researchers did not break down their age groups sufficiently, making it difficult to determine the exact age at which children were acquiring different features of language variation.

More recently, Jacewicz and Fox (2019) found that 9–12-year-old children speaking Appalachian English increased the degree of diphthongization of /aɪ/ in their reading style, while their pronunciation was more monophthongal (and thus closer to the local regional dialect form) in casual style. They also noted that girls used the diphthongal realization more than boys in reading style but not during interviews. Likewise, Kaiser (2022) observed that 3–6-year-old Austrian children used more dialect features in play mode than when storytelling. Such variation was also constrained by the interlocutor, whereby the youngsters used more standard features when retelling a narrative to a standard speaker than they did when the addressee was using dialect, further indicating children’s ability to adapt their speech to various factors including interlocutor and setting.

However, children’s ability to vary along a stylistic continuum is constrained by their linguistic repertoires (Andersen 1992), which, in the case of diglossic style-shifting especially, is further complicated by exposure to and ability in SA. Such competence varies both at the abstract linguistic level and in performance. Moreover, it is subject to various factors including education, attitudes, and motivation (Mejdell 2006, p. 2).

A consequence of the diglossic situation in Arabic contexts is that SA is not acquired natively by children, with consistent and sustained exposure only coming through education around school age (Ferguson 1959, 1991; Mejdell 2006; Saiegh-Haddad 2012). Given the differences between SA and VA, both structurally and at the level of the lexicon, diglossic style-shifting is, therefore, expected to be challenging for young children. For example, Saiegh-Haddad (2012) finds that for five-year-old children, there is only a 20% overlap between SA and VA lexical items. Conversely, 40% of their lexicon is comprised of uniquely VA items, i.e., vocabulary which does not occur in SA. The remaining 40% consists of ‘paired lexical items.’ Importantly, the latter often differ phonologically either on the segmental or syllable structure level or both. For example, SA (θaldʒ) ‘snow’ is realized as [talʒ], [θalidʒ], or [taliʒ] in different spoken vernaculars. In other words, it differs on a segmental level in the first realization, on the syllable structure level in the second, and on both in the last. On the other hand, Albirini (2016, p. 33) maintains that despite being sporadic and informal, children’s exposure to SA actually begins very early in life through various means. These include cartoons and other forms of media; learning the Quran; and reading or being read children’s books, which enables them to develop receptive skills in SA in a similar vein to those who might be classified as unbalanced bilinguals. Similarly, Sabir and Safi (2008) observe various instances of diglossic CS in a five-year-old Saudi boy, including his use of SA verbs, verb phrases, and lexical items, as well as phonological features. Evidence of early diglossic awareness and receptive skills in SA were also found by Leikin et al. (2014), who examined the narrative ability of five-year-old Palestinian children. Although more successful in VA story retelling, children in their study were able to use some SA linguistic structure and showed an ability to understand narratives when delivered in SA.

3.2. Gender

In traditional accounts of style, intraspeaker variation is reflective of interspeaker variation (Hernández-Campoy and Cutillas-Espinosa 2012). As such, female speakers are typically more likely to style-shift and ‘hypercorrect’ in the direction of standard and prestigious variants than male speakers, given the association of such variants with higher social classes (e.g., Labov 1966; Romaine 2008; Trudgill 1972). In later accounts of stylistic variation, style is used to construct and project different identities, including gendered identities, by using linguistic variables which create rather than simply reflect social meaning (e.g., Eckert 1988, 2003). In the context of Arabic, this is further complicated by the diglossic situation, which leads to two levels of competing prestige within the same speech community (Abd-El-Jawad 1987; Ibrahim 1986).

SA derives much of its prestige from being the only written variety with a codified orthography and well-defined rules, making it the ‘uncontested prescriptive norm for reading and writing in Arabic-speaking communities’ (Mejdell 2006, p. 2). This contributes to its high literary value as well as associations with education, learnedness, and formality (Albirini 2016; Mitchell 1986) and gives it values of ‘eloquence’ and ‘correctness’, as reflected in the label used for it in Arabic1 (Ibrahim 1986; Maamouri 1998). SA is also the official language in all Arabic-speaking countries and is the language of government and administration, leading to additional associations with authority and social mobility (Albirini 2016). Moreover, being the language of the Quran and the Islamic literary tradition has given SA a high level of religious prestige to the point of being perceived by many as a ‘sacred’ language, rather like Latin once was in Western contexts (Albirini 2016; Ferguson 1968; Versteegh 2010). As such, SA has always enjoyed a higher status than the spoken vernaculars, which are sometimes viewed as ‘corruptions’ of it (Maamouri 1998). However, the spoken vernaculars do have their own hierarchy of prestige independently from SA and along a continuum of rural, urban, and Bedouin dialects (Abd-El-Jawad 1987; Ibrahim 1986; Mitchell 1986). So, unlike the contexts surrounding most contemporary European languages, the notions ‘standard’ and ‘prestige’ cannot be used interchangeably in studies of Arabic sociolinguistics (Abd-El-Jawad 1987). In fact, proximity to the standard plays no role in the prestige of spoken varieties, but rather the status of a dialect and its speech community (Abd-El-Jawad 1987; Holes 1995). Indeed, urban varieties are deemed the most prestigious in many Arabic-speaking communities despite their divergence from SA (Habib 2010; Miller 2004), and speakers of peripheral dialects that share features with SA may abandon these features in favor of urban variants that may, in fact, be distant from the standard (e.g., Al-Ali and Arafa 2010; Amara 2005; Habib 2010, 2011). For example, Habib (2010, 2011) reports a preference by speakers of a rural dialect in Syria for the urban [ʔ] over the local [q], despite the latter’s overlap with the standard. Similarly, Amara (2005) finds that speakers in Bethlehem are abandoning their local [θ] for the urban [t], although the latter diverges from the standard. While preference for the overtly prestigious urban variants is mostly shown by female speakers, such choices are also sometimes reported for males and linked to place of residence, desire for social mobility, or even as an index of a supralocal identity (e.g., Al-Wer and Herin 2011; Habib 2010).

While SA may not play an active role in variation and change on the level of spoken Arabic as the above studies suggest (Al-Wer et al. 2022), its presence in the Arabic sociolinguistic landscape, as well as the status it enjoys, makes it an important stylistic resource for Arabic speakers. Early studies of Arabic sociolinguistic patterns assumed a preference by male speakers for SA variants as an index of access to power, authority, and public life (e.g., Bakir 1986; Sallam 1980). Although these assumptions were refuted for failing to account for the two levels of prestige that exist in Arabic, as well as not disentangling local from standard variants in cases of overlap, the assumption that male speakers favor SA variants more than their female peers still persists (Chakrani 2015; Daher 1998; Miller 2005). Miller (2005), who investigated rural migrants in Cairo, bases her conclusions on a higher frequency of lexical borrowing and use of standard forms in the speech of male speakers in her sample. Different results come from Bassiouney (2009), who examined the use of SA in the speech of highly educated men and women hosting four talk shows in an Egyptian context. Use of SA appeared to be constrained by function, conveying a sense of assertiveness and finality, and denoting an identity of power and authority. Her results show that women have the same access to SA as men and, in some cases, they use it even more.

4. Social and Linguistic Background of the Study

The current study was carried out in Khan Eshieh Camp, a community of Palestinian refugees that was established in 1949 about 25 km to the southwest of Damascus. Three major varieties of Arabic contribute to the rich sociolinguistic landscape of Khan Eshieh: the local dialect, Damascene Arabic, and Standard Arabic. The local dialect of the speech community is a Bedouin dialect and shares many features with Bedouin dialects across the Levant (see, e.g., Palva 2006; Rosenhouse 1982). Relevant to this study, these features include the realization of the standard (q) as a voiced velar stop [ɡ], as in [ɡaɫam] ‘pen’ for (qalam), and the retention of standard interdental fricatives, (θ), (ð). Exposure to Damascene Arabic comes through dialect contact in the form of geographical diffusion (Britain 2002), given Khan Eshieh’s proximity to Damascus as well as the latter’s status as a major urban center in Syria, and the status of its dialect as the national standard and the variety’s prevalence in the media (Miller 2004). Additionally, Khan Eshieh’s population is highly integrated in Syria and many community members are well educated and very active in the Syrian labor market, indicating a high level of mobility among camp residents. Damascene Arabic is an urban dialect and shares many features with other major urban dialects in the Levant. These include the realization of the standard (q) as a glottal stop, as in [Ɂalam] ‘pen’ for (qalam), and the realization of dental fricatives as either stops ([t] as in [tuːm] ‘garlic’ for (θuːm) and [d] as in [dahab] ‘gold’ for (ðahab) or alveolar fricatives ([s] as in [masalan] ‘for example’ for (maθalan) and [z] as in [laziːz] ‘delicious’ for (laðiːð) (Al-Wer 2007; Al-Wer and Herin 2011). Exposure to SA mainly happens through education, but informal experiences of the variety also stem from both the media and participating in religious practices.

5. Materials and Methods

This research is part of a larger project investigating the acquisition of sociolinguistic variation by Arabic-speaking children and adolescents. It aims to examine children’s and adolescents’ language development in a linguistically diverse environment and to investigate how they navigate their different linguistic resources in socially meaningful ways.

5.1. Participants

Forty boys and girls, aged between 3;7–17;9, were recruited for the study using a snowball sampling technique. All of them were born and raised in the speech community, as were their parents. Participants and their families were contacted in person or via telephone and prior informed consent was obtained from participants and their parents. They were also provided with age-appropriate project descriptions prior to data collection.

Participants were divided into 5 age groups corresponding to well-defined stages in the educational system in Syria, since school has a central role in the life of children and adolescents and the formation of their social networks, which is expected to have a significant impact on their linguistic behavior and language use (Eckert 2017). A similar age division is found in Habib (2011, 2014, 2016, 2017), who examined patterns of variation and change in the speech of rural Syrian children and adolescents. Participants were further subdivided by gender, as

Table 1. Participant groups.

Age Group	3–5	6–8	9–11	12–14	15–17
Boys	4	4	4	3	4
Girls	6	4	3	4	4

below shows. Children between the ages of 6and 14, corresponding to grades 1 through 9, attended school in 6 separate groups that were segregated by gender due to issues with space in the local schools. The youngest age group (3–5-year-olds) were preschoolers and the oldest (15–17-year-olds) were in high school (grades 10–12). Unlike the 6–14-year-olds, participants in the oldest age group attended a mixed school for both boys and girls.

5.2. Data Collection

Following the Labovian paradigm of style investigation (Labov 1972, 1981), participants’ speech was recorded in two different settings varying in perceived formality. Participants’ spontaneous productions were recorded during semi-structured sociolinguistic interviews (Labov 1972; Tagliamonte 2006) as they responded to open-ended questions about their hobbies, daily routines, leisure activities, school experiences, and community life. A set of open-ended questions, loosely based on Tagliamonte (2006), was prepared for use in the interviews (see Appendix A). However, these served only as prompts, with priority given to topics that emerged as more productive and elicited personal narratives to encourage participants to lead the conversation and reduce the observer’s paradox (Labov 1972). Variation in context was achieved using a picture-naming task, which was expected to invoke a school setting and thus prompt participants to use standard realizations signaling diglossic style-shifting in their speech. A full set of the tokens used for the picture-naming task can be found in Appendix B. Data collection was carried out by a local female fieldworker who was likewise born and raised in the community, so there would be minimal expected impact of the interlocutor’s dialect on the participants’ language use. She was 25 years old at the time of the recording, which allowed her to connect quite well with participants in the oldest group. She flattered the male speakers by calling them “young men,” inquiring about their outings, and engaging enthusiastically in discussions about their future. With the girls, she emphasized their shared experiences, saying things like “we girls should change the status quo” when they expressed frustrations with the patriarchal society.

Traditional variationist studies under the Labovian paradigm use a reading passage, word list, and a minimal pair list alongside a sociolinguistic interview in order to increase the formality of the speech event (e.g., Labov 1972; Trudgill 1972). These could not be used in the current study, as the sample included very young children who still could not read. More importantly, reading tasks are not appropriate for a study on Arabic, since they would automatically introduce SA, it being the main written variety (Al-Wer et al. 2022). Therefore, it was necessary to use a different elicitation technique that would obtain specific linguistic information on the use of the linguistic variables under investigation. Picture-naming has been used in studies on acquisition and linguistic development in both Arabic and English (e.g., Amayreh and Dyson 1998; Dyson and Amayreh 2000; Foulkes et al. 2001). It has also been employed in sociolinguistic studies with adult speakers (e.g., Taqi 2010 on Kuwaiti Arabic speakers) when investigating style-shifting (e.g., Sandow 2022 on Cornish speakers). Crystal (1976) observes that phonological markers are among the most discernible of style-shifting, so the task was expected to produce fruitful results for tracking stylistic variation. However, given the overlap between SA and some of the local realizations, other indications of variation in response to the context will likewise be considered.

A list of digital pictures of familiar animals and household objects was used for the picture-naming task. A pilot test of the task was carried out beforehand to determine that the pictures were appropriate, especially for the youngest participants, by administering it to a 3-year-old to ensure that even that cohort could readily follow and engage with the task. Each picture was shown to the child on a slideshow, and he readily named them without prompting. Token items were chosen to feature the variables of interest in different word positions where possible. Eighteen tokens were used to elicit (q) variants, six tokens to elicit (θ), and five to elicit (ð) variants.

The picture task was run on a slideshow displayed on a laptop and each child was asked to name the item in question. As expected, the task was found to exert a level of formality, and an association with school was obvious in the responses of participants, as will become clearer in the discussion of the results.

Data were transcribed orthographically in ELAN 4.7.1 (Wittenburg et al. 2006), an annotation software for audio and video that provides several valuable tools for transcribing linguistic data (Brugman and Russel 2004). The software enables the segmentation of audio files into manageable utterances for transcription and includes a playback feature. Additionally, ELAN’s search capabilities allow for the identification and playback of individual variables (Brugman and Russel 2004). Manual auditory coding followed Labov’s (1972) principle of accountability, where every possible instance of the variable was documented. Percentages were then calculated for each variant based on its frequency relative to all potential occurrences in each task. For instance, if there were 100 environments for (q) in the interview context and [q] occurred 10 times, this would constitute 10% of the (q) realizations in that context, and so on. Realizations of the variables were quantified and coded according to speakers’ gender and age group in addition to the nature of the two tasks, i.e., either the sociolinguistic interview or the picture-naming task. A total of 716 tokens were elicited for (θ), 435 for (ð), and 1895 for (q).

6. Results

The data analysis in this study is primarily quantitative, as is typical for variationist research. However, this is complemented by some qualitative analysis in the discussion to obtain a more rounded understanding from a social perspective of the diglossic style-shifting patterns observed (particularly important given the overlap between standard and local variants of the interdental fricatives already described).

The data analysis was carried out using SPSS 25.0 (The Statistical Package for the Social Sciences). Style-shifting across the picture-naming task and the interview contexts was examined using a paired-samples t test, as it allows for a statistical comparison of mean values across tasks (Griffith 2010). In the following sections, results on overall style-shifting patterns are presented, followed by a breakdown of the results by age, gender, and the interaction of the two.

6.1. Style-Shifting in the Use of (θ)

As noted in the introduction above, interdental fricatives are retained in Bedouin dialects and, as such, overlap with standard realizations. Urban realizations, i.e., stops and sibilants, are also present in the speech community under study as a result of influences from Damascene Arabic, as explained in Section 4 above. However, the use of the urban variants is expected to be less frequent in the picture-naming task for the reasons already stated.

Indeed, despite the overlap between local and standard realizations, use of [θ] was still more frequent in the picture-naming task, whereas use of the urban variant [t] was more prevalent in the interview context, as

Table 2. Distribution of (θ) variants across contexts.

Context	[θ]	%	[t]	%	[s]	%	Total
Interview	378	77.1%	97	19.8%	15	3.1%	490
Picture task	185	81.9%	37	16.4%	4	1.8%	226

and Figure 1 below indicate. The urban fricative [s] was used highly sporadically in both contexts, but it still surfaced more in the interview than in the picture-naming task. The differences in using [θ] and [t] across contexts were found to be highly significant (p = 0.002 for [θ] and p = 0.001 for [t]).

6.1.1. The Influence of Age on Style-Shifting Patterns in the Use of (θ)

Greater use of [θ] in the picture task appeared in the speech of all groups. Little difference was recorded in the speech of the 15–17-year-old group, who used the variant categorically in the picture task and near-categorically in the interview context. The most notable differences appeared in the speech of 3–5, 9–11, and 12–14-year-old speakers, as

Table 3. Distribution of (θ) variants across contexts by age group.

Age Group	Context	[θ]	%	[t]	%	[s]	%	Total
3–5	Interview	17	48.6%	18	51.4%	0	0%	35
	PT	26	54.2%	18	37.5%	4	8.3%	48
6–8	Interview	32	72.7%	12	27.3%	0	0%	44
	PT	37	75.5%	12	24.5%	0	0%	49
9–11	Interview	96	76.8%	23	18.4%	6	4.8%	125
	PT	38	95%	2	5%	0	0%	40
12–14	Interview	67	57.8%	42	36.2%	7	6%	116
	PT	36	87.8%	5	12.2%	0	0%	41
15–17	Interview	166	97.6%	2	1.2%	2	1.2%	170
	PT	48	100%	0	0%	0	0%	48

and Figure 2 below illustrate. These differences were significant for 3–5-year-old speakers, whereby use of the standard variant [θ] was significantly higher in the picture task at p = 0.016, while the urban variant [t] was significantly favored in the interview context: p = 0.002. This is especially remarkable considering their young age and presumed inexperience with SA. Significant differences were also observed in the speech of 12–14-year-olds, who utilized the standard variant significantly more in the picture task at p = 0.048, while they used the urban variant [t] significantly more in the interview context: p = 0.030. The greater use of the latter during the interview by 12–14-year-old speakers was mainly due to the preference for urban variants by female speakers in the group, as will become evident in Section 6.1.3, Section 6.2.3 and Section 6.3.3 (see Shetewi 2018, 2023 for further details).

6.1.2. The Influence of Gender on Style-Shifting Patterns in the Use of (θ) Variants

Both male and female speakers used [θ] more in the picture task than in the interview context, as is evident in

Table 4. Distribution of (θ) variants across contexts by gender.

Gender	Context	[θ]	%	[t]	%	[s]	%	Total
Male	Interview	202	82.4%	37	15.1%	6	2.4%	245
	PT	90	81.8%	18	16.4%	2	1.8%	110
Female	Interview	176	71.8%	60	24.5%	9	3.7%	245
	PT	95	81.9%	19	16.4%	2	1.7%	116

and Figure 3 below. Differences were significant in the speech of male speakers, who used the standard variant significantly more in the picture task (M = 81.58, SD = 28.77) than in the interview context (M = 65.56, SD = 41.67) at p = 0.008. Their use of the urban variant [t] was significantly less frequent in the same context: p = 0.006. For female speakers, differences in using the standard variant in the picture task (M = 78.64, SD = 30.71) and the interview context (M = 64.68, SD = 33.80) were not significant at p = 0.059. However, their preference for the urban variant was significantly less in the picture task at p = 0.045. It is also worth noting that while female speakers used [θ] noticeably less than their male peers during interviews, this difference disappeared in the picture-naming task, suggesting style-shifting among female speakers in more formal contexts. This outcome also indicates that males and females are equally adept at shifting across a range of styles including SA when the context requires it.

6.1.3. The Influence of the Interaction of Age and Gender on Style-Shifting Patterns in the Use of (θ) Variants

As

Table 5. Distribution of (θ) variants across contexts by age and gender.

Male Speakers
Age Group	Context	[θ]	%	[t]	%	[s]	%	Total
3–5	Interview	1	7.1%	13	92.9%	0	0%	14
	PT	9	39.1%	12	52.2%	2	9%	23
6–8	Interview	5	41.7%	7	58.3%	0	0%	12
	PT	18	78.3%	5	21.1%	0	0%	23
9–11	Interview	88	86.3%	8	7.8%	6	5.9%	102
	PT	23	100%	0	0%	0	0%	23
12–14	Interview	47	83.9%	9	16.1%	0	0%	56
	PT	16	94.1%	1	5.9%	0	0%	17
15–17	Interview	61	100%	0	0%	0	0%	61
	PT	24	100%	0	0%	0	0%	24
Female Speakers
3–5	Interview	16	76.2%	5	23%	0	0%	21
	PT	17	68%	5	24%	2	8%	25
6–8	Interview	27	84.4%	5	15.6%	0	0%	32
	PT	19	73.1%	7	26.9%	0	0%	26
9–11	Interview	8	34.8%	15	65.2%	0	0%	23
	PT	15	88.2%	2	11.8%	0	0%	17
12–14	Interview	20	33.3%	33	55%	7	11.7%	60
	PT	20	83.3%	4	16.7%	0	0%	24
15–17	Interview	105	96.3%	2	1.8%	2	1.8%	109
	PT	24	100%	0	0%	0	0%	24

and Figure 4 below show, most speakers, as predicted, used the standard variant [θ] more in the picture task than they did in the interview context. This was especially noticeable in the speech of 3–5-year-old boys as well as 9–11 and 12–14-year-old girls. Significant differences occurred amongst the cohort of 3–5-year-old boys, who favored the standard variant significantly more in the picture task at p = 0.021 and thus used the urban variant [t] significantly less in this context: p < 0.001. Significant differences also occurred in the speech of 12–14-year-old girls, whose preference for the standard variant was significantly higher in the picture task at p = 0.026 than in the interview context, where their use of the urban variant [t] was significantly greater: p = 0.004. Some interesting exceptions did, however, occur. Girls in the 3–5- and 6–8-year-old cohorts, for example, used the standard variant [θ] less in the picture task than might have been expected.

6.2. Style-Shifting in the Use of (ð)

This section examines stylistic variation in the use of (ð) variants. As already noted, similar to (θ) above, there is an overlap between the standard and local realizations of this variable, too. The standard variant [ð] was still used more frequently in the picture task despite the overlap. On the other hand, [d] was indeed favored in the interview context, as illustrated in

Table 6. Distribution of (ð) variants across contexts.

Context	[ð]	%	[d]	%	[z]	%	Total
Interview	167	69.6%	71	29.6%	1	0.4%	240
PT	156	80%	36	18.5%	1	0.5%	326

and Figure 5 below. A paired-samples t test revealed the difference in using the variants to be highly significant at p = 0.006 for [ð] and p = 0.010 for [d].

6.2.1. The Influence of Age on Style-Shifting Patterns in the Use of (ð)

The most noticeable stylistic variation in the realization of the variable occurred in the speech of 6–8, 9–11, and 12–14-year-old speakers, as shown in

Table 7. Distribution of (ð) variants across contexts by age group.

Age Group	Context	[ð]	%	[d]	%	[z]	%	Total
3–5	Interview	17	40.5%	24	57.1%	0	0%	42
	PT	19	44.2%	21	48.8%	1	2.3%	43
6–8	Interview	27	50.9%	26	49.1%	0	0%	53
	PT	32	76.2%	10	23.8%	0	0%	42
9–11	Interview	47	78.3%	13	21.7%	0	0%	60
	PT	33	94.3%	2	5.7%	0	0%	35
12–14	Interview	22	71%	8	25.8%	1	3.2%	31
	PT	33	91.7%	3	8.3%	0	0%	36
15–17	Interview	54	100%	0	0%	0	0%	54
	PT	39	100%	0	0%	0	0%	39

and Figure 6 below. However, significant differences only appeared in the speech of 9–11-year-olds, whose use of [ð] was significantly higher in the picture task than in the interview context at p = 0.032.

6.2.2. The Influence of Gender on Patterns of Style-Shifting in the Use of (ð)

Female speakers used the standard variant significantly more in the picture task than in the interview context: p = 0.008. However, males’ use of the variant was slightly lower in the picture task. This surprising pattern is likely due to the overlap between standard and local realizations of (ð) and males’ general preference for the variant in both contexts, as is evident from

Table 8. Distribution of (ð) variants across contexts by gender.

Gender	Context	[ð]	%	[d]	%	[z]	%	Total
Male	Interview	101	82.8%	21	17.2%	0	0%	122
	PT	77	77.8%	21	21.2%	0	0%	99
Female	Interview	66	55.9%	50	42.4%	1	0.8%	118
	PT	79	82.3%	15	15.6%	1	1%	96

and Figure 7 below. Unlike boys, who used the variant predominantly in both contexts, girls’ use of the variant was relatively infrequent in the interview context, making the difference in their use of it across contexts statistically significant. Both males and females use the variant similarly in the picture task, which is another indication of style-shifting in the speech of female participants. As noted in Section 6.1.2 above, these results also suggest that while there were marked gender differences observable in the interview context which shows female participants’ preference for urban variants in more informal settings, such differences were less marked in the picture task. This would appear to indicate that gender is not a reliable predictor of when standard features will be used in this community.

6.2.3. Register Variation and (ð) Variants: The Interaction Between Age and Gender

Table 9. Distribution of (ð) variants across contexts by age and gender.

Male Speakers
Age Group	Context	[ð]	%	[d]	%	[z]	%	Total
3–5	Interview	3	21.4%	11	78.6%	0	0%	14
	PT	6	38.6%	14	66.7%	0	0%	21
6–8	Interview	9	60%	5	40%	0	0%	15
	PT	16	76.2%	5	23.8%	0	0%	21
9–11	Interview	45	93.8%	3	6.3%	0	0%	48
	PT	20	100%	0	0%	0	0%	20
12–14	Interview	17	94.4%	1	5.6%	0	0%	18
	PT	16	88.9%	2	11.1%	0	0%	18
15–17	Interview	27	100%	0	0	0	0%	27
	PT	19	100%	0	0	0	0%	19
Female Speakers
3–5	Interview	14	50%	14	50%	0	0	282
	PT	13	59.1%	7	31.8%	1	4.5%	223
6–8	Interview	18	47.4%	20	52.6%	0	0	38
	PT	16	76.2%	5	23.8%	0	0	21
9–11	Interview	2	16.7%	10	83.3%	0	0	12
	PT	13	86.7%	2	13.3%	0	0	15
12–14	Interview	5	38.5%	7	53.8%	1	7.7%	13
	PT	16	88.9%	2	11.1%	0	0	18
15–17	Interview	27	100%	0	0	0	0	27
	PT	20	100%	0	0	0	0	20

and Figure 8 below show that the use of [ð] was greater in the picture-naming task than in the interview context for most speakers. Noticeable differences appeared in the speech of male and female speakers in the 6–8-year-old group, as well as among female speakers in the 9–11 and 12–14-year-old cohorts. For example, use of the variant rose from 38.5% in the interview context to 94.4% in the picture task amongst 12–14-year-old girls. Participants in the 15–17-year-old group used the variant categorically in both contexts, so that no variation in the realization of the variable could be observed in their speech production.

Even though differences in the realization of the variable were not significant for any particular group, girls’ general tendency to favor the use of urban variants gives a clear indication that SA plays a role in their linguistic choices, contrary to the assumption that SA has a bigger influence on the speech of males in Arabic-speaking communities more widely (e.g., Amara 2005; Chakrani 2015; Daher 1998; Miller 2005).

6.3. Style-Shifting in the Use of (q)

This section examines stylistic variation in speakers’ realizations of (q). Quantitative analysis of such variation across contexts is actually more straightforward in the case of (q), since it does not overlap with any dialectal variants in the community, i.e., the local [ɡ] or the urban [Ɂ]. However, given that the variable is highly prone to lexical conditioning and borrowing from SA (Al-Wer and Herin 2011; Habib 2010; Holes 1995), the contexts for such realizations will also be examined to establish a more comprehensive account of diglossic style-shifting in the speech of these young participants. Independent coding was carried out for lexical conditioning. Tokens invariably realized with [q] are excluded from the quantitative analysis of style-shifting and are instead examined in Section 6.3.4 below.

As hypothesized, use of the standard variant [q] was substantially greater in the picture task than in the interview context. It should be noted that none of these instances were due to lexical conditioning, as the items chosen for (q) in the picture task would all have [g] realizations in the local dialect. A paired-samples t test revealed the difference in frequency to be highly significant at p < 0.001. In turn, the local variant was used significantly less in the picture task at p < 0.001. Use of the urban variant was significantly less in the picture task as well, at p = 0.022. Despite the considerable increase in using the standard variant in the latter, the local variant was still the most frequent, as can be seen from

Table 10. Distribution of (q) variants across contexts.

Context	[q]	%	[ɡ]	%	[Ɂ]	%	Total
Interview	143	12.1%	941	79.8%	94	8%	1179
Picture task	281	39.2%	374	52.2%	51	7.1%	716

and Figure 9 below.

6.3.1. The Influence of Age on Style-Shifting Patterns in the Use of (q)

As

Table 11. Significant differences in using the standard and local variants across contexts by age group.

Age Group	[q]	[ɡ]
3–5	p = 0.003 *	not significant
6–8	p = 0.003 *	p = 0.003 *
9–11	p < 0.001 **	p = 0.002 *
12–14	p = 0.011	not significant
15–17	p = 0.037	p = 0.039

A single asterisk refers to statistically significant results, double asterisks refer to results which are highly significant.

below shows, speakers in all age groups used the standard variant significantly more in the picture task than they did in the interview context, with the most drastic differences appearing in the speech of 6–8, 9–11, and 12–14-year-old speakers. The youngest children, the 3–5-year-olds, exhibited an impressive ability to style-shift given their limited exposure to SA. In the same vein, the use of the local variant in the picture task was significantly less for most age groups too. Noticeable differences in using the urban variant only appeared in the speech of the 12–14-year-old cohort, but these were not significant.

Despite the considerable and statistically significant increase in the participants’ use of the standard variant in the picture-naming task, the local variant was still the most frequent in the speech of most groups in both contexts, as

Table 12. Distribution of (q) variants across contexts by age group.

Age Group	Context	[q]	%	[ɡ]	%	[Ɂ]	%	Total
3–5	Interview	7	8.2%	65	76.5%	12	14.1%	85
	PT	38	24.7%	90	58.4%	16	10.4%	154
6–8	Interview	21	12.4%	127	74.7%	22	12.9%	170
	PT	81	49.1%	62	37.6%	22	13.3%	165
9–11	Interview	41	11.7%	298	84.9%	12	3.4%	351
	PT	68	51.9%	61	46.6%	2	1.5%	131
12–14	Interview	35	16.4%	131	61.5%	47	22.1%	213
	PT	57	45.2%	58	46%	11	8.7%	126
15–17	Interview	39	10.8%	320	88.9%	1	0.03%	360
	PT	37	26.4%	103	73.6%	0	0%	140

and Figure 10 below show.

6.3.2. The Influence of Gender on Style-Shifting Patterns in the Use of (q)

Greater use of the standard variant in the picture task appeared in the speech of all participants, as

Table 13. Distribution of (q) variants across contexts by gender.

Gender	Context	[q]	%	[ɡ]	%	[Ɂ]	%	Total
Male	Interview	85	12.2%	555	79.6%	57	8.2%	697
	PT	131	37.1%	181	51.3%	35	9.9%	353
Female	Interview	58	12%	386	80.1%	37	7.7%	482
	PT	150	41.3%	193	53.2%	16	4.4%	386

and Figure 11 below show. Similarly, use of the local variant decreased in the speech of both male and female speakers in this context. A paired-samples t test revealed that differences in the frequencies of the local variant [ɡ] and the standard variant [q] were highly significant at p < 0.001. For female speakers, significant differences only occurred in their use of the standard variant at p < 0.001. However, there were no differences in their choices for the local variant.

6.3.3. The Influence of the Interaction of Age and Gender on Style-Shifting Patterns in the Use of (q)

Table 14. Distribution of (q) variants across contexts by age and gender.

Male Speakers
Age Group	Context	[q]	%	[ɡ]	%	[Ɂ]	%	Total
3–5	Interview	2	4.4%	34	75.6%	9	20%	45
	PT	13	18.3%	39	54.9%	13	18.3%	71
6–8	Interview	6	12%	32	64%	12	24%	50
	PT	35	49.3%	22	31%	14	19.7%	71
9–11	Interview	21	7.4%	254	89.4%	9	3.2%	284
	PT	40	50.6%	39	49.4%	0	0%	79
12–14	Interview	29	17.7%	108	65.9%	27	16.5%	164
	PT	21	33.9%	33	53.2%	8	12.9%	62
15–17	Interview	27	17.5%	127	82.5%	0	0%	154
	PT	22	31.4%	48	68.6%	0	0%	70
Female Speakers
3–5	Interview	5	12.5%	31	77.5%	3	7.5%	40
	PT	25	30.1%	51	61.4%	3	3.6%	83
6–8	Interview	15	12.5%	95	79.2%	10	8.3%	120
	PT	46	48.9%	40	42.6%	8	8.5%	94
9–11	Interview	20	29.9%	44	15.7%	3	4.5%	67
	PT	28	53.8%	22	42.3%	2	3.8%	52
12–14	Interview	6	12.12%	23	46.9%	20	40.8%	49
	PT	36	56.3%	25	39.1%	3	4.7%	64
15–17	Interview	12	5.8%	193	93.7%	1	0.5%	206
	PT	15	21.4%	55	78.6%	0	0	70

and Figure 12 collate the usage of (q) variants across contexts by age and gender. As can be seen, speakers in all groups exhibited a greater use of the standard variant in the picture task, while their preference for the local variant decreased in this context. Significant differences in using the variants were observed in most speaker groups. Males in the youngest cohort used the local variant [ɡ] significantly less in the picture task than they did during the interview at p = 0.035, while girls of the same age preferred the standard variant significantly more in the picture task at p = 0.020. Use of the standard variant in the picture task was also significantly greater in the speech of 6–8-year-old female speakers at p = 0.012, while use of the local variant [ɡ] was significantly less at p = 0.009. Male speakers in the 9–11-year-old group also used the standard variant significantly more in the picture task at p = 0.003, while the local variant [ɡ] was significantly less utilized at p = 0.028. However, no such significant distinctions appeared for girls in the same cohort. In addition, there was no difference in the frequency of use of the standard variant in the speech of 12–14-year-old boys at p = 0.065, but a highly significant difference was observed for girls in the same group: p = 0.018. By contrast, there were no significant differences whatsoever found in the occurrence of the standard versus local variants in the speech of 15–17-year-old participants.

6.3.4. Lexical Conditioning in Realizations of (q)

The variable (q) is known to be resistant to variation in contexts involving lexical borrowing from SA, such as quoting the Holy Quran, regardless of speakers’ age, gender, or dialect background (Al-Wer and Herin 2011; Habib 2010; Holes 1995; Miller 2005). Use of [q] is thus reported to be limited to these contexts when there is no overlap with any dialectal variants4 (Al-Wer and Herin 2011; Cotter 2016; Miller 2005; Ornaghi 2010). However, this proved not to be the case in the current study. Invariable realizations with [q] only made up 4.2% of all realizations of the variable, and 16.7% of all realizations with [q], suggesting that 83.3% of [q] realizations were actually not lexically conditioned at all and instead appeared to compete with the other two dialectal variants, namely, the local [ɡ] and the urban [Ɂ]. The majority of such realizations (66.3%) occurred in the picture-naming task, as we have seen above, suggesting a strong influence of context on speakers’ choices. The remaining 33.7% appeared in the interview context, which suggests that SA has a considerable influence on speakers’ linguistic behavior.

This section will focus on the categorical use of [q] in the interview, which made up 6.7% of all realizations of (q) in this style and 37.3% of all realizations of [q] in the same context. These realizations are conjectured in the current study to stem from (i) borrowing from SA and (ii) words that may be realized with dialectal variants in the traditional dialect or other dialects but were consistently realized with [q] in our data.

The first category included standard lexical items, literary and technical words, and Quran and literary quotations. The use of [q] in these contexts, which is well attested in the literature (Albirini 2016; Davies and Bentahila 2008; Miller 2005; Ornaghi 2010), occurred in the speech of most participants but was most evident in that of the oldest group as they talked about their hobbies and future, or in social and political commentary. A 17-year-old girl, for example, expressing her frustration with traditional gender roles and what she views as males’ sense of entitlement, quoted a verse from the Quran that she felt was overly misquoted by men to justify such entitlement:

(1)	a.	dʊɣɾi	j.iɡʊl-ik	‘ʔ-aɾɾɪdʒæ:lʊ	qɑwwæ:mu:na		ʕala	n-nisæ:ʔ’5
		immediately	3S.say-you	‘t men	maintainers		on	the women
		They immediately tell you ‘Men are the maintainers of the women’.
	b.	smiʕt-ha	ʃi:	χamis	maɾɾæt		hɑi	i-ssina
		Heard-it	some	five	times		this	year
		I heard it about five times this year.
(2)	a.	il-ha	maʕna	θæːni	ʔin-ʊ	t.iɾʕɑ:-ha		tihtam
		For-it	meaning	another	it-3S	2S-take care-her		care for
		It has another meaning, to take care of her, care for her.
	b.	fi:-ha	mu:	qɑwwɑːm	ʕ ali:-ha
		in-her	not	responsible	on-her
		Not be in charge of her

These examples exhibit an interesting pattern of variation in using (q). The speaker, for example, uses the local variant [ɡ] in jiɡʊli-k ‘tell you’ outside of the verse but still uses the standard variant for qɑwwɑːm ‘maintainer’, even when it is part of her own speech and not in the verse itself, since the word is borrowed from SA and, as such, is lexically conditioned.

Another example comes from the speech of a different girl in the group who quoted a famous saying about pride and confidence when asked if she has a Facebook account in her own name:6

(3)	e:h	lakæ:n!	‘wɑːθiqʊ	lχʊtˤwɑti	yamʃi:	malakan!’
	Yes	of course	confident	of step	walks	king
	Yes, of course! ‘He who’s with confidence, walks like a king!’

She also used items such as [muʃɑwwiqa] ‘exciting’ and [munɑmmɑqa] ‘sophisticated’, talking about her passion for reading and how it impacted upon her vocabulary.

Another quotation of a learned saying came when discussing the security situation at the time of the interviews. It occurred in the speech of a 17;3-year-old male who expressed his acceptance of the risk, saying that if he were to die on his way to school, he would die a martyr. His utterance included other lexical items that were realized with [q] as well:

(4)	a.	hijja	ɾu:ħ	wiħda	w	qɑdɑɾ
		It	soul	one	and	destiny
		We only live once, and everything is destined.
	b.	bas	ʔ.ʕɾif	miːn	zat-ha	l-qɑðiːfa
		only	1s.know	who	threw-it	the-shell
		But I want to know who fired the shell.
	c.	bas	j.imuːt	ʃahiːd
		but	3S.die	martyr
		But one dies a martyr.

(5)	a.	‘man	mæːta	fiː	tˤɑɾiːqi-hi	ʔila	l-ʕɑmɑl	fa	hʊwwa	ʃahiːd
		who	dies	in	way-his	to	the-work	so	he	martyr
		‘The one who dies going to work is a martyr…
	b.	w	man	mæːta	ʕalæː	maqʕadi	d-dɪɾæsa	fa	hʊwwa	ʃahiːd’7
		so	he	martyr	and	who	dies
		and the one who dies while in school is a martyr.’

In addition to the first category, items that may be realized with the dialectal variants in other varieties or even in the traditional dialect were categorically or near-categorically realized with [q] in the data. For example, words like [tɑqri:ban] ‘approximately’, [ɾɑqɑm] ‘number’, [fɑɾiːq] ‘team’, [qɑsˤif] ‘bombing’, and [qɑðiːfa] ‘shell’ were realized with [q] consistently in the data (c.f., Cotter 2016; Habib 2010 who reported many of these items as realized with the urban [ʔ]). These items comprised most of invariable realizations with [q] and some of them occurred even in the speech of the youngest groups. For example, a five-year-old girl told the interviewer that she was not attending kindergarten because it was hit by a shell:

(6)	bɑtˤtˤɑl.na	n.ɾu:ħ	mni	tˤ-tˤɑχtˤɑχɑ	wi	d-dab,	ʒai	qɑði:fa	hnæ:k
	stopped.1P	1P.go	from	the-shooting	and	the-bombing	came	shell	there
	We stopped going because of the shooting and bombing, a shell hit there!

Abd-El-Jawad and Suleiman (1990) report many of these items as realized with [q] and propose lexical conditioning as an explanation for such realizations. A more plausible explanation is that although these items are not exclusively standard, they share the same phonological structure as the SA forms, minimizing their phonological distance from the standard to one phoneme only, in this case the standard /q/ (Saiegh-Haddad 2003; Saiegh-Haddad and Haj 2018). This phonological relatedness to the standard is likely why [q] is used in their realization.

6.3.5. Lexical Conditioning in Relation to Age

Although age did not greatly impact the combined use of [q], when considered independently, both categorical and optional use of [q] were, in fact, influenced by age. Figure 13, where the two categories are calculated out of all the tokens realized with [q] in the interview context, illustrates that categorical use of the variant rises in linear fashion with age and is visibly greater in the speech of older speakers. By contrast, use of [q] in optional contexts decreases with age and is lowest in the speech of the oldest group.

This is unsurprising, since the use of [q] in obligatory contexts mostly occurred in Quran quotations, learned sayings, and lexical borrowing from SA. Such usage would naturally be expected to be greater in the speech of older speakers, whose linguistic repertoires are richer than those of younger speakers because they have had more time in education and thus better access to this register of language and a broader knowledge base. On the other hand, the use of the variant in free variation was higher in the speech of younger speakers. It could be argued that younger speakers are more influenced by school than older speakers and would therefore use the standard variant of (q) even in everyday words. This is reflected in comments from participants and their caregivers and further enhanced by the community at large. For example, children are routinely asked about their school year and how well they are doing in school, and parents often comment on their children’s educational achievements as either a signal of praise or rebuke. Education is highly regarded in the community, which is instilled in children from a young age. This is likely because most members of the community feel disenfranchised as refugees with little else in the way of social mobility and economic prosperity.8 For older speakers, although education continues to enjoy a high regard, the local dialect is the primary spoken form, and they reserve their use of the standard to appropriate topics and contexts.

6.3.6. The Influence of Gender on the Lexical Conditioning of (q)

Use of the standard variant in obligatory and optional contexts was relatively comparable in the speech of both male and female speakers, although its categorical use was somewhat greater in the speech of girls. The use of the variant in the free variation was, on the other hand, more frequent in the speech of the males, as is evident from Figure 14 below. These results clearly show that gender had, in fact, no influence on using the standard variant of (q).

7. Discussion

This paper presented an analysis of diglossic style-shifting in the speech of Arabic-speaking children and adolescents. The research was undertaken utilizing the Labovian paradigm with a modified methodology to fit the Arabic context. As discussed in Section 3 above, while most Western studies use reading tasks to examine style-shifting (Labov 1984), this, as already stated, is impractical in an Arabic setting where writing is traditionally linked to SA. Additionally, reading tasks are not suitable for the youngest pre-literate or early literacy age group. Instead, a picture-naming task was used, since it was expected to invoke a formal school setting, prompting speakers to shift toward SA, which is associated with education. Indeed, Fischer (1958) suggested that children associate a school setting with formality, leading them to switch to a formal register. As already noted, African American children used Standard English during classroom activities, whereas they used slang in informal situations (De Stefano 1972; Melmed 1973). The results of our study show that the picture task did, indeed, introduce a level of formality invoking a school setting. For example, a speaker in the 15–17-year-old group asked whether he should ‘read’ the pictures, and the use of standard variants was significantly higher when naming items from the pictures than in the sociolinguistic interview. Moreover, features readily identified with the standard were reserved for picture-naming, while, interestingly, speakers still used the vernacular when conversing with the fieldworker during the task. This indicates that, despite their age, these youngsters had a high level of awareness of relevant contexts and appropriate speech styles.

Quantitative results reveal stylistic variation in the realization of all variables under examination, despite the overlap between local and standard realizations of the interdental fricatives. Given such overlap and the overwhelming preference for local variants by certain speaker groups during the interview task, it is important to examine other features that may denote the use of the standard and not limit the discussion to statistically significant results. Speakers employed a variety of SA features during the picture-naming task, which, in addition to the statistically significant results detailed in the previous sections, indicate a high level of diglossic style-shifting in their speech. For example, speakers resorted to the use of uniquely standard lexical items in place of dialect words, such as [miðˤɑlla] ‘umbrella’ rather than [ʃamsijja]. With cognate words, speakers used the standard vocalic structure in place of the vernacular, such as in [ðubæ:ba] ‘fly’ rather than the colloquial [ðibbæ:na] and [ðail] ‘tail’ in place of the local [ðe:l]. This was the most common strategy, and the most discernible, especially in cases of overlap between the standard and dialectal variants of the variables under study. Other examples include [θawɾ] ‘bull’ in place of the local [θo:ɾ], [baidˤ] ‘eggs’ in place of the local [be:ðˤ] or urban [be:dˤ], and [ħaʊdˤ] ‘tank’ rather that the local [ħo:ðˤ] or urban [ħo:dˤ]. This occurred even in the speech of a five-year-old girl, who used the standard vocalic structure in [ħimɑ:ɾ] ‘donkey’ rather than the vernacular [ħmɑ:ɾ].9 This is remarkable, especially in the speech of such a young, inexperienced speaker, since matching the vocalic structure of the standard is reported to be amongst the most difficult aspects of acquisition in a dialect–standard continuum context (Saiegh-Haddad 2003; Saiegh-Haddad and Haj 2018). In some instances, partially standard phrases were used in the task. For example, a 10-year-old boy responded with [be:t maksuw biθθaldʒ] ‘a house covered in snow’ for the target /θaldʒ/, making use of the standard lexical item [maksʊw] rather than simply responding with the local [θalidʒ]. Here as well, despite the word for ‘snow’ being nearly identical in both SA and the local dialect, the standard syllabic structure was used. In addition to these features, a 14-year-old boy jokingly used tanween-, a grammatical feature that is mostly exclusive to SA, alongside other features in his responses, as in [qalamon] ‘pen’ and [ðaɪlon] ‘tail’. Although the feature was used jokingly, it indicates both awareness of the relevant speech style and skill in using it appropriately.

As already discussed, statistically significant variation based on context occurred even in cases of overlap between the standard and local variants, as in the case of the interdental fricatives. In this regard, style variation was most noticeable in the speech of girls between the ages of six and fourteen as use of the urban variants in the interview context was greatest in the speech of these groups, making their switch to the standard realization readily identifiable.

Examination of stylistic variation was more straightforward in the case of (q), since its standard variant does not overlap with any dialectal variants relevant to the community. Use of the variant would therefore be safely assumed to approximate the standard. Miller (2005) similarly notes that the use of standard [q] is independent of dialectal variation when the varieties involved do not have it as a native variant. Statistical analysis showed that use of [q] was significantly higher in the picture-naming task, indicating a clear effect of perceived formality on the choice of linguistic variants. Several super tokens (Tagliamonte 2012) occurred in the realization of the variable in the task. For example, [qalam ʔazɾaɡ] and [ɡaɫam ʔazɾaq] ‘blue pen’ occurred in the speech of many speakers. Others provided multiple realizations of the same target word. A 17-year-old boy, for example, responded with [qaʊs, qo:s, ɡo:s] ‘headband’. Note that in the first response, the vocalic structure of Arabic was used, whereas in the second, only the standard variant of (q) was employed. Speakers’ awareness of the task’s formality and use of appropriate features to express it were also evident when certain individuals realized some of the same words differently when they occurred in the interview context and when the same items were fortuitously repeated in the picture task. Two male speakers in the 9–11-year-old group, for example, used the local variant in (baqaɾa) ‘cow’ when it occurred in the interview context, but used the standard variant in the picture task. In fact, it is interesting that variation in the use of (q) between the picture task and interview context occurred in the speech of all participants but was mostly noticeable in speakers between the ages of 12 and 14. Although gender differences did appear in the interview data, with a higher preference for urban variants shown by female speakers (especially in the 9–11 and 12–14-year-old cohorts), gender proved not to be indexed by this particular type of variation, which contravenes previous assumptions that Arab men use standard forms more than women do (Daher 1998; Chakrani 2015; Miller 2005). Miller (2005, p. 933), who studied rural migrants in Cairo, generalizes her findings to all Arabic-speaking communities, claiming that men, regardless of level of education, tend to use more standard features than their female peers do. The results of this study, however, show that standard forms actually occur at noticeably frequent rates even in the speech of 9–11 and 12–14-year-old girls, who strongly favor urban features. Use of [q] in the interviews as a function of topic, discussed further in Section 6.3.4 above, also shows no differences between male and female speakers. The frequency with which the variant was produced in both the picture task and the sociolinguistic interview presents an interesting pattern that seems to be highly dependent on age. Recall that the use of the [q] variant, which occurred in free variation, was greater in the speech of younger speakers, whereas [q] in lexically conditioned environments was produced more often by the oldest cohort. The influence of school and its role in shaping linguistic practices may be greater in the speech of younger groups, which may explain their use of the standard variant in free variation with the local one. Such influences may, in fact, be deduced through speakers’ own comments during the data collection process. For example, a five-year-old girl who could not readily remember the word for ‘padlock’ (qifl) was prompted by her mother, who used the standard variant for (q), adding ‘remember when we learnt this for ‘qaf’?’, also referring to the variable as a letter in the alphabet.

These examples indicate children’s familiarity with the standard variety and its influence on their speech. Indeed, although the diglossic situation is hypothesized to negatively impact certain skills such as reading acquisition, it is found to improve metalinguistic awareness (Eviatar and Ibrahim 2001). However, the latter alone is not enough without linguistic competence, as it only implies a passive control of register (Andersen 1992). Linguistic competence, which is highly variable among Arabic speakers, as noted earlier, is governed by factors such as age, education, attitude, motivation, and so forth (Mejdell 2006). As such, proficiency in SA is an important consideration when examining switches between SA and VA, as some speakers may lack the proficiency to consistently use SA in conversation (Hudson 2002). This explains why speakers’ use of [q] based on topics and in technical lexical items is most frequent in the speech of the oldest group. Speakers in that group have the linguistic repertoire necessary to discuss such topics with an appropriate style, whereas younger speakers still lack such competence despite their awareness of the appropriate style. In addition, the older speakers, based on their age, experience, and education, have more interest than younger speakers in discussing topics dealing with religion and politics, which have been shown to invoke the use of standard features in previous research (e.g., Albirini 2011; Miller 2005). Nonetheless, results on diglossic style-shifting in the speech of these young participants indicate an impressive level of mastery both in terms of competence and performance given the distance between SA and VA, which renders learning SA to be almost like acquiring a foreign language (Ibrahim and Aharon-Peretz 2005; Saiegh-Haddad 2004).

Results on style variation, especially those relating to (q), also suggest that SA has a powerful influence on the speech of children and adolescents in the community. This outcome may also indicate that SA has a stronger impact than the urban dialect does on their speech. Together with education, the spread of all-day channels that offer cartoons dubbed in SA may have helped in spreading standard features—especially in the case of young speakers. For older participants, their use of SA may serve as an intermediate form between features that are highly local and those that are urban, since SA is considered a shared register between all Arabic speakers that transcends dialects and geography (Ferguson 1959). This is a tentative assumption in light of the findings observed here. Such sentiments were, however, expressed by some adult members10 of the community, who said that they would rather make use of standard features than switch instead to markedly urban ones.

8. Conclusions

The results of this study exhibit variation between three forms of Arabic in the speech community, with participants availing of local, urban, and standard features in their repertoire. The results on diglossic style-shifting, which locate the picture-naming task, in addition to topics that prompt lexical borrowing from SA, as the main sites for using standard features, further exemplify the diglossic nature of Arabic, since both local and urban features co-occur in the same domain. Moreover, this study’s findings echo those of Shetewi (2018, 2023), since they reveal that even very young members of this community possess multilectal repertoires which they navigate strategically to make socially meaningful linguistic choices. Such competencies are already attested to in the speech of educated adult speakers (e.g., Albirini and Chakrani 2017), though this is the first, as far as we are aware, to find evidence for this type of sophisticated sociolinguistic awareness among young Arabic speakers. Indeed, as Sandow (2022, p. 680) observes, being aware of the social meaning of different linguistic variants and the ability to negotiate and draw on them appropriately demonstrates speakers’ competence as ‘social actors’.