Accommodation Patterns in the Speech of Arabic-Speaking Children and Adolescents: A Variationist Analysis

Abstract

This paper presents a variationist analysis of patterns of speech accommodation by 40 Arabic-speaking children and adolescents (aged 3–17) experiencing dialect contact in a Bedouin speech community near Damascus, Syria. It examines participants’ use of the phonological variables (θ), (ð), and (q), and the morphophonological feminine suffix (-a) in recorded sociolinguistic interviews and play sessions with two female fieldworkers, a local and an urban speaker, in order to investigate accommodation patterns across different interlocutors. Accommodation patterns were influenced by age, gender, and the linguistic variable under examination. Convergence to the urban interviewer was most evident in the realization of (q), whereas little convergence, and indeed variation, occurred in the realization of (-a), and more convergence occurred in the speech of girls and speakers younger than 15. Divergence and maintenance emerged in the speech of 15–17-year-old male speakers. These patterns are analysed in light of Accommodation Communication Theory and issues of identity and linguistic prestige in Arabic. Accommodative behaviour in the speech of participants exhibits their awareness of the social value of the phonological variables under investigation and demonstrates a high level of sociolinguistic awareness and competence.

1. Introduction

Speech accommodation, which refers to altering one’s speech in response to an interlocutor (Giles et al. 1991; Gallois et al. 2005; Dragojevic et al. 2016), is a hallmark of human interaction. Gasiorek et al. (2015) assert that it is part of human nature, and Britain and Trudgill (2009) suggest that it occurs as a natural outcome of dialect contact between speakers of mutually intelligible varieties and becomes a driving force for language variation and change (Pardo 2006; Garrett and Johnson 2013). Speech accommodation has been examined and attested across different languages and in different contexts (Giles et al. 2023). Such accommodation can be manifested as convergence, divergence, or maintenance whereby speakers may adapt their language use to the interlocutor, distance it from the interlocutor, or maintain their own speech patterns, respectively (Giles et al. 1991; Gallois et al. 2005; Dragojevic et al. 2016; Giles 2016). Speakers may accommodate to their interlocutors through a variety of linguistic features including lexical items, syntactic forms, or phonological variants, among others (Giles et al. 1987, 1991). This would involve adaptation of familiar features over which speakers already have control when accommodation occurs within a speech community but would entail learning new forms in situations of new or unfamiliar varieties (Trudgill 1986; Dragojevic et al. 2016). As such, speakers’ linguistic knowledge and repertoires, which can be dependent on several factors including age, education, and exposure to certain linguistic features (Andersen 1992), would play a key role in the degree and patterns of accommodation (Dragojevic et al. 2016; Gasiorek et al. 2022). Other factors are also projected to influence accommodation including similarities and/or differences between the varieties in contact (Hernández 2002); underlying beliefs and attitudes towards both the interlocutor and the linguistic forms of choice in each interaction (Giles et al. 1991; Gasiorek and Giles 2012); identity perceptions (Giles et al. 1991; Miller 2005); and linguistic prestige (e.g., Miller 2005; Giles and Ogay 2007; Habib 2010). For example, when contact occurs between high prestige and low prestige varieties, speakers of stigmatized dialects may view convergence to the prestigious norms as a necessity for social mobility and integration (Miller 2005; Habib 2010). Likewise, speakers converge to individuals they respect and admire or to those they associate with a socially attractive group (Giles et al. 1991; Giles 2008, 2016). On the other hand, identity perceptions may prompt divergent behaviour whereby speakers accentuate their own speech forms to disassociate themselves from the interlocutor, which, in turn, would enhance their identity and their sense of self (Giles et al. 1991; Gasiorek 2016; Zhang and Giles 2018). Convergence and divergence can also have long-lasting effects on speech communities, especially in situations of dialect contact. Long-term societal and personal convergence may contribute to language change (Trudgill 1986; Giles et al. 1991), while long-term intergroup divergence could contribute to language maintenance and survival (Giles et al. 1991). Auer and Hinskens (2005, pp. 335–36) propose that for interpersonal accommodation to lead to language change, three hierarchically ordered components must be in place. The first relates to the ‘interactional’ episode whereby speakers may demonstrate short-term accommodation by converging to their interlocutor; either by adopting the interlocutor’s, usually innovative or prestigious, forms, or by abandoning their own, usually traditional, features. The second relates to the speaker who would demonstrate long-term accommodation by adopting such features in their own speech (i.e., outside of the interactional episode). The third component relates to the speech community at large whereby such innovations are spread into the community.

As such, while it can be argued that accommodative behaviour, especially phonetic convergence, maybe an automatic reflex (e.g., Goldinger 1998; Delvaux and Soquet 2007), patterns of speech accommodation are impacted by a number of socio-cognitive factors including age, gender, linguistic knowledge, linguistic prestige, and linguistic attitudes (Giles et al. 1991; Dragojevic et al. 2016). Indeed, Hinskens et al. (2005) assert that accommodative acts are conscious choices of socially aware individuals.

2. Accommodation Patterns in the Speech of Children and Adolescents

2.1. Age

Despite being influenced by the factors stated in the introduction above, research shows that accommodative linguistic behaviour emerges in children as early as 2–3 years of age (e.g., Andersen 1984; Lanza 1992; Youssef 1993; Paugh 2005; Montanari 2009). However, little is known about patterns of speech accommodation by children and adolescents in the Middle East and North Africa. This paper aims to fill this gap by examining the occurrence of cross dialectal accommodation in the speech of Arabic-speaking children and adolescents experiencing dialect contact. Results from this paper will add to the existing literature on accommodation theory and shed light on its applicability in a non-Western context. Research of this nature is essential as it contributes to the understanding of children’s linguistic development and their language attitudes, as well as shedding light on the linguistic trends in the region.

Early studies of child language were largely predicated on the idea that children were monostylistic speakers with full awareness of the social functions of language presumed to only appear in adolescence (Labov 1964; Lakoff 1973). Ample evidence, however, shows that children as young as two or three years old can use their language appropriately depending on context and formality, as well as accommodate their speech to the communicative needs of different interlocutors (e.g., Gleason 1973; Andersen 1984; Youssef 1993; Paugh 2005; Montanari 2009). Some early evidence comes from Street and Cappella (1989), who examined accommodation patterns in 3–6-year-old children to an adult female in dyadic interviews and reported that they accommodated her in pauses, turn taking, and speech rate. More recently, Kaiser (2022, p. 50) examined accommodation patterns in the speech of 3–6-year-old Austrian children and found that they adapt their use of standard vs. dialect according to the interlocutor. Results from the study also found that the children use more dialect features in play vs. storytelling. Indeed, linguistic variation is acquired as part of the linguistic system and appears in the speech of children as early as age two or three (Andersen 1992; Kiesling 2011; Roberts 2013). Research shows early acquisition of stylistic constraints on speech alongside, and in some cases prior to, acquisition of grammatical constraints (e.g., Foulkes et al. 2001; Chambers 2009; Johnson and White 2020). As such, children develop early awareness of the various linguistic resources available in their environment and often make strategic and socially meaningful use of such resources. For example, Khattab (2013) finds that 5-year-old English-Arabic bilinguals acquire a variety of linguistic forms available in their environment, including their parents’ non-native accented forms, and use this varied repertoire and manifest it in convergence and divergence strategies in different communicative situations. She notes that accommodation mechanisms are essentially the same for bilingual, monolingual, and bidialectal children (Khattab 2013, p. 469). Habib (2016, 2017) finds that variation in the use of rural vs. urban features by rural Syrian children and adolescents is influenced by both age and gender, whereby boys abandon urban features at the age of eight, whereas girls retain them into adolescence. She argues that such patterns result from different associations with these features and that they contribute to the creation of highly differentiated gendered linguistic behaviour, which indicates early development of sociolinguistic competence on the part of those young speakers. Children, therefore, have the appropriate sociolinguistic knowledge to utilize whatever linguistic forms they have at their disposal for effective communication with different interlocutors in varying situations.

While an ability for linguistic accommodation appears in children as young as 3 or 4 years old, such skills develop with age as the sociolinguistic knowledge and ability to control the cognitive, social, and psychological mechanisms that determine the degree and level of linguistic accommodation also develop with age (Leaper 1991; Youssef 1993; Dossey et al. 2020). Andersen (1992) notes that the speakers’ knowledge of style might be passive or active depending on their linguistic competence, and that since children do not have access to the full range of styles in language (Kerswill 1996), it is expected that their ability to accommodate would develop as their linguistic competence develops. Babel (2009) proposes that socially motivated accommodation may develop with age as children’s social and psychological abilities mature, while cognitive automatic stimuli for speech accommodation start as early as the babbling stage.

In dialect contact situations, like the one examined in this current study, the ability to accommodate would also require knowledge of the varieties involved (Hernández 2002; Dragojevic et al. 2016). Research finds that early exposure to second dialect features is essential for native-like acquisition (e.g., Chambers 1992; Hernández 2002; Starks and Bayard 2002), and that young children are more likely to adopt new linguistic forms as a desire to fit in with their peers and belong in their new community (Chambers 2002).

2.2. Gender

Ample empirical evidence in various areas of language use has shown clear patterns of gender-based linguistic practices. While this is not to be understood as indicating essentialist differences between speakers of different genders, it does exemplify performances of gender identities in relevant communities and contexts.

In terms of accommodation, research finds that speakers of all genders may choose to converge to their interlocutors, especially in mixed-sex interactions, to reduce gendered language differences (Robertson and Murachver 2003, p. 321; Levitan et al. 2012; Demina 2021). For example, Al-Wer and Herin (2011) report that young male speakers’ use of the urban variant [ʔ] as a realization of (q) increases when they converse with young women. Conversely, speakers may wish to diverge and accentuate gendered language differences in order to achieve distance from the interlocutor. For example, Levon and Holmes-Elliott (2013) report that working-class female speakers accentuate sex differences in their articulation of /s/, especially in mix-sex interactions in order to construct a class-based feminine identity. However, in very general terms, research finds that female speakers tend to converge more than male speakers, especially in mixed-sex interactions (Giles and Ogay 2007; Lelong and Bailly 2011; Palomares et al. 2016; Perez-Sabater 2017). Namy et al. (2002, p. 23) also note that both male and female speakers accommodate less to female interlocutors. Female speakers also tend to be more innovative in their language use and more likely to adopt incoming linguistic features than male speakers (Labov 2001). As such, in a context of cross-dialectal accommodation such as the one examined in this paper, it is expected that female speakers will be more accommodative than male speakers.

Gender differences in accommodation patterns and strategies start emerging at an early age since gender differences in language use appear quite early and increase with age (Sheldon 1990; Robertson and Murachver 2003; Hilte et al. 2022). Indeed, Habib (2011b, 2014, 2017 on rural Syrian children) finds that boys increase their use of local, rural features after the age of 8, whereas girls persist in using urban features. Similarly, Shetewi (2018 on Palestinian children) finds that boys increase their use of local Bedouin features after the age of 5, with sharp gender differentiation most apparent in children between the ages of 6 and 14. As children get older, they start converging to the other sex in mixed interactions and their accommodation strategies become similar to those of adults (Robertson and Murachver 2003, p. 323). For example, in a study on teenagers’ social media writing, Hilte et al. (2022, p. 252) find that boys use more oral markers (e.g., use of colloquial terms) than girls in any setting, but that such use decreases in mixed-gender interactions, indicating convergent behaviour on their behalf. A similar pattern emerges for girls who use a less ‘female’ style when interacting with boys. Such accommodative behaviour in mixed-sex interactions is reported even before school age (Leaper 1991; Killen and Naigles 1995). Gender differences in accommodation appear even in same-sex interactions both in terms of the level and patterns of accommodation. For example, in a study of same-sex peer interactions of 11–15-year-old African American children, Van Hofwegen (2015, pp. 37–38) finds that girls are generally more accommodative than boys and that they are more likely to converge to strangers than to friends, whereas boys tend to diverge with unfamiliar interlocutors but converge with friends. Boys and girls were also found to accommodate different linguistic features whereby girls were highly accommodative with ethnically salient features, regardless of the interlocutor, whereas boys did not accommodate such features with unfamiliar interlocutors (Van Hofwegen 2015, pp. 39–41), which may suggest that girls are more concerned about belonging and drawing on a shared identity than boys.

It is worth noting that although gender differences play a role in how accommodation is manifested, when gender and style are experimentally controlled, speech style appears to play a bigger role in controlling accommodative acts for all speakers regardless of gender (Hannah and Murachver 1999; Thomson and Moore 1999; Thomson et al. 2001), which also applies to children. For example, speech style has more of an effect on the accommodation patterns of 6–11-year-old children than the gender of either the child or the interlocutor (Robertson and Murachver 2003). Boys with a strong masculine identity were still found to be much less likely to accommodate to female-gendered speech, however (Robertson and Murachver 2003). This is argued to be a result of boys having a stronger sense of in-group identity than girls and may feel socially threatened if they are perceived to converge to female speech (Leaper 2000). Still, Van Hofwegen (2015, p. 31) claims that unless gender is salient in an interaction, it does not play much of a role in the linguistic choices of speakers. However, as gender and constructing gender identities is of key importance to adolescents (Eckert 2005), it is important to consider it in accommodation studies (Van Hofwegen 2015).

In the community under study in this paper, girls and boys go to separate schools between the ages of 6 and 14 years, which may focus their peer groups to within their own sex. They start forming their own games, even outside of school, which is presumed to further contribute to conformity to gender norms (Golombok et al. 2008). Such games mostly conform to gender stereotypes as girls are normally not allowed to play outside as much as boys. Boys in the study, for example, reported playing ‘thieves and police’ and various other forms of tag, going on their bikes, and generally playing out on the street, whereas girls mostly reported playing house and tea parties and watching popular TV series. Both reported not wanting to play with the other sex at this age and girls reported going on mostly female visits with their mothers as a favourite activity. This is expected to impact their language use both in terms of accommodation and variation. As such, girls are expected to converge more than boys, especially given their preference for prestigious, urban forms (e.g., Habib 2014, 2016, 2017).

3. Social and Linguistic Background of the Study

The study was carried out in Khan Eshieh Camp ((χæ:n iʃ-ʃi:ħ), a community of Palestinian refugees that was established in 1949 about 25 km to the south-west of Damascus, shown in Figure 1 (Google Maps 2023). Khan Eshieh’s population is well integrated in Syria and many of them are highly educated and active in the Syrian labour market, which indicates a high level of mobility among camp residents. This, in addition to geographical proximity to Damascus, leads to dialect contact with Damascene Arabic. Given the status of Damascus as a major urban centre in Syria and the status of its dialect as the national standard, as well as its prevalence in the media (Miller 2004), such dialect contact is expected to be in the form of geographical diffusion (Britain 2002) of urban forms into nearby localities, including the speech community under investigation.

The heritage dialect of Khan Eshieh is a traditional Bedouin dialect and shares many features with Bedouin dialects in the Levant (see, e.g., Rosenhouse 1982; Palva 2006). Relevant to this paper, these features include the realization of the standard (q) as a voiced velar stop [ɡ] as in [ɡaɫam] ‘pen’ for (qalam), the retention of standard interdental fricatives, (θ), (ð), and no raising of the feminine marker (-a). Damascene Arabic is an urban dialect and shares many features of major urban dialects in the Levant, including the realization of the standard (q) as a glottal stop as in [ʔalam] ‘pen’ for (qalam), the realization of interdental fricatives as stops ([t] as in [tu:m] ‘garlic’ for (θu:m) and [d] as in [dahab] ‘gold’ for (ðahab)) or alveolar fricatives ([s] as in [masalan] ‘for example’ for (maθalan) and [z] as in [lazi:z] ‘delicious’ for (laði:ð), and conditional raising of the morphophonemic feminine suffix (-a) to [e], as in [waɾde] ‘flower’ for (waɾda) (Al-Wer 2007; Lentin 2007; Al-Wer and Herin 2011; Habib 2011a). Importantly, in the context of the Levant, urban realizations are associated with prestige, social power, and mobility, despite their divergence from the standard (Al-Wer 2003; Amara 2005; Al-Wer and Herin 2011). Bedouin features, in contrast, are reported as isolated minority features that are usually abandoned in favour of the urban variants, especially in the speech of young women who are often found to favour overtly prestigious urban features (Al-Wer 1991; Amara 2005; Al-Ali and Arafa 2010).

4. Materials and Methods

This research is part of a larger project investigating acquisition of sociolinguistic variation by Arabic-speaking children and adolescents. The choice of the speaker sample is motivated by an interest in children’s and adolescents’ acquisition of variation in dialect contact settings and aims to uncover how they acquire and make use of the different linguistic resources available to them in these situations.

4.1. Participants

Employing a snowball sampling technique, forty boys and girls, aged 3; 7–17; 9, with no known speech or language delays were recruited to participate in the study. They were all born and raised in the speech community to parents who were also native to the community. Their exposure to urban features came primarily from the media and from adult speakers whose speech had become variable thanks to their mobility in the direction of Damascus for work and education (Shetewi 2018), as well as from urban teachers who taught in the local schools. Contact with the participants and their families was made in person or via telephone and informed consent was obtained from both participants and their parents. Age-appropriate project descriptions were also provided to participants and their parents 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 behaviour and language use (Eckert 2017). A similar age division is also found in Habib (2011b, 2014, 2016, 2017), who examined patterns of variation and change in the speech of rural Syrian children and adolescents. Participants were further divided 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 6 and 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 the pre-schoolers 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.

4.2. Data Collection

Data collection was carried out by two female fieldworkers who lived in the community and were known to most participants and to all their parents. The first fieldworker is a native of the community, which is reflected in her speech and her realizations of the variables of interest in this study. She was born and raised in the community and was 25 at the time of data collection. The second fieldworker is married into the community and had been teaching at the local high school for about 17 years at the time of recording. She was 58 years old at the time of data collection and had been living in the community for 29 years. Although she speaks what can be classified as an urban variety, interdental fricatives are overwhelmingly retained in her speech, which has interesting implications on participants’ patterns of accommodation, as will be explained in the discussion. This may be a result of prolonged contact with the community or due to her being a teacher and regularly reciting the Quran, so that her productions are more standard-like, which overlaps with the local variants. Her realization of (-a) where conditioning allows is categorically urban and her realization of (q) is also categorically urban, apart from instances of lexical borrowings from Standard Arabic. It should be noted here that many of the participants’ parents were former students of the urban fieldworker. Additionally, in a small, close-knit community with dense multiplex networks (Milroy 1986), many of the participants, especially in the oldest group, were familiar with her either in her capacity as a teacher (of parents, older siblings, or cousins, etc.) or through family relationships. This familiarity, as well as her status, may impact accommodation patterns in the speech of adolescent boys and girls (e.g., Van Hofwegen 2015).

Semi-structured sociolinguistic interviews (Labov 1972; Tagliamonte 2006) and play sessions with open-ended questions about hobbies, daily activities, pastime activities, school life, and life in the community were recorded by the two fieldworkers, in consecutive sessions of 30–45 min, in order to elicit participants’ spontaneous speech and allow for an examination of accommodation patterns across the different interviewers. More specifically, interviews were started by the local fieldworker in order to elicit participants’ most common realizations when interacting with community members and then taken over by the urban fieldworker to introduce a different, more prestigious variety and examine whether that would trigger speech accommodation on the part of participants. The single-session recordings were carried out in the homes of participants, who were interviewed individually in the presence of a parent or caregiver (mostly mothers).

Data were transcribed orthographically in ELAN (Wittenburg et al. 2006) and manual auditory coding was performed according to Labov’s principle of accountability (1972). Realizations of the variables were quantified and coded according to speakers’ gender and age groups in addition to the two interview contexts: (i) the interview with the Bedouin interlocutor; and (ii) the interview with the urban interlocutor. A total of 816 tokens were elicited for (θ), 396 tokens were elicited for (ð), 1807 tokens were elicited for (q), and 1786 tokens were elicited for (-a).

5. Results

This study is a variationist analysis of speech accommodation and is, therefore, primarily quantitative, following the Labovian paradigm. Some qualitative analysis is still included in the discussion, however, to complement the quantitative results and further enhance the discussion of accommodation in the speech of participants. As noted above, coding of the data was performed according to Labov’s principle of accountability (1972). That is, every possible occurrence of a token was recorded for each context. Given that the data analysed here are primarily spontaneous speech, the number of tokens for any given variable varied greatly amongst speakers. Therefore, percentages for variant distribution were calculated for each variant out of all possible occurrences in each context. For example, if 100 environments for (q) occurred in the local interview context and [ɡ] was used 35 times, it would represent 35% of (q) realizations in this context, and so forth.

Data analysis was carried out using SPSS 25.0 (The Statistical Package for the Social Sciences). Accommodation across the urban and local 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 accommodation patterns are presented followed by a breakdown of the results by age, gender, and the interaction of age and gender. Overall accommodation patterns are presented, followed by a breakdown of these results by age and gender.

5.1. Accommodation Patterns in the Use of (θ)

As noted in the introduction above, interdental fricatives are retained in Bedouin dialects like that of the speech community but are realized as alveolar stops or sibilants in urban Levantine dialects.1 Given their overt prestige (Amara 2005; Al-Ali and Arafa 2010), it was expected that use of the urban variants would be higher in the interview with the urban interviewer. Indeed, although the local [θ] was the majority variant in both interviews, speakers’ use of the urban variants was noticeably higher in the urban interview context. A paired-samples t test revealed that these differences were highly significant. Speakers used the local variant [θ] significantly less with the urban interviewer than they did with the local interviewer: p = 0.003. They used the urban stop variant [t] significantly more in the interview with the urban speaker at p = 0.009. The urban sibilant variant was also used significantly more in the interview with the urban speaker despite its overall sporadic use in the data: p = 0.030.

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

Interviewer	[θ]	%	[t]	%	[s]	%	Total
Local	378	77.1%	97	19.8%	15	3.1%	490
Urban	177	54.3%	128	39.3%	21	6.4%	326

and Figure 2 below demonstrate the use of (θ) variants across interview contexts. Here and throughout, the labels ‘local’ and ‘urban’ represent the speech variety of the interviewers. The y-axis shows variant frequencies and the colour of the columns represents the different variants: blue for local realizations and orange and grey for urban realizations.

5.1.1. The Influence of Age on Accommodation Patterns in the Use of (θ)

Accommodation towards the urban speakers occurred in varying degrees in the speech of all age groups. It was most noticeable in the speech of 3–5, 9–11, and 12–14-year-olds, with the urban variant [t] being the majority variant in the speech of these groups in the urban interview context, as evident from

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

Age Group	Interviewer	[θ]	%	[t]	%	[s]	%	Total
3–5	Local	17	48.6%	18	51.4%	0	0	35
	Urban	9	33.3%	18	66.7%	0	0	27
6–8	Local	32	72.7%	12	27.3%	0	0	44
	Urban	25	59.5%	17	40.5%	0	0	42
9–11	Local	96	76.8%	23	18.4%	6	4.8%	125
	Urban	38	40.9%	48	51.6%	7	7.5%	93
12–14	Local	67	57.8%	42	36.2%	7	6%	116
	Urban	29	37.7%	40	51.9%	8	10.4%	77
15–17	Local	166	97.6%	2	1.2%	2	1.2%	170
	Urban	76	87.4%	5	5.7%	6	6.9%	87

and Figure 3 below. Significant differences in using the local variant [θ] across interviews appeared in the speech of 9–11-year-old speakers who used the variant significantly less in the interview with the urban interlocutor: p = 0.036.

5.1.2. The Influence of Gender on Accommodation Patterns in the Use of (θ)

Both male and female speakers accommodated their realizations of (θ) in the interview with the urban speaker, as evident from

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

Gender	Interviewer	[θ]	%	[t]	%	[s]	%	Total
Male	Local	202	82.4%	37	15.1%	6	2%	245
	Urban	94	67.1%	41	29.3%	5	4%	140
Female	Local	176	71.8%	60	24.5%	9	3.7%	245
	Urban	83	44.6%	87	46.8%	16	8.6%	186

and Figure 4 below. Such accommodation was only statistically significant in the speech of female speakers, however. Their use of the urban [t] increased significantly in the urban interview context at p < 0.001, while their use of the local variant decreased significantly at p < 0.001.

5.1.3. The Influence of Age and Gender on Accommodation Patterns in the Use of (θ)

Most speakers in most groups accommodated their speech towards the urban interviewer, as seen in

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

Male Speakers
Age Group	Interviewer	[θ]	%	[t]	%	[s]	%	Total
3–5	Local	1	7.1%	13	92.9%	0	0	14
	Urban	3	25%	9	75%	0	0	12
6–8	Local	5	41.7%	7	58.3%	0	0	12
	Urban	11	55%	9	45%	0	0	20
9–11	Local	88	86.3%	8	7.8%	6	5.9%	102
	Urban	30	68.2%	9	20.5%	5	11.4%	44
12–14	Local	47	83.9%	9	16.1%	0	0	56
	Urban	16	55.2%	13	44.8%	0	0	29
15–17	Local	61	100%	0	0	0	0	61
	Urban	34	97.1%	1	2.9%	0	0	35
Female Speakers
3–5	Local	16	76.2%	5	23%	0	0	21
	Urban	6	40%	9	60%	0	0	15
6–8	Local	27	84.4%	5	15.6%	0	0	32
	Urban	14	63.6%	8	36.4%	0	0	22
9–11	Local	8	34.8%	15	65.2%	0	0	23
	Urban	8	16.3%	39	79.6%	2	4.1%	49
12–14	Local	20	33.3%	33	55%	7	11.7%	60
	Urban	13	27.1%	27	56.3%	8	15.7%	48
15–17	Local	105	96.3%	2	1.8%	2	1.8%	109
	Urban	42	80.8%	4	7.7%	6	11.5%	52

and Figure 5 below. This was negligible in the case of male speakers in the 15–17-year-old group, who used the local variant overwhelmingly in both interview contexts. Another exception applied to boys in the two youngest groups who used the local variant slightly more in the interview with the urban interlocutor. The general pattern of accommodation remained in the direction of urban realizations, however. This was most evident in the speech of female speakers in the 3–5 and 6–8-year-old cohorts and was revealed to be significant in the speech of both groups at p = 0.029 for 3–5-year-old girls and p = 0.018 for 6–8-year-old girls. Considerable accommodation also seemed to occur in the speech of 9–11 and 12–14-year-old boys. Girls in the 9–11 and 12–14-year-old groups were found to strongly favour the urban variants in both interview contexts. As such, their convergence to the urban interlocutor was not found to be statistically significant. Indeed, girls in the 9–11-year-old group used the local variant significantly less than their male peers in both interview contexts (p = 0.017 in the local interview context and p = 0.015 in the urban interview context).

5.2. Accommodation Patterns in the Use of (ð)

As expected, the use of the local variant [ð] was higher in the local interview context, while the urban variants [d] and [z], despite the sporadic use of [z], were used more in the urban interview context. The differences were not statistically significant, but an obvious trend of convergence towards the urban speaker is evident in the speech of participants, as

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

Interviewer	[ð]	%	[d]	%	[z]	%	Total
Local	167	69.6%	71	29.6%	1	0.4%	240
Urban	85	54.5%	65	41.7%	5	3.2%	156

and Figure 6 demonstrate. Indeed, their use of the local variant decreased from about 70% in the local interview context to a little over 50% in the urban interview context. On the other hand, their use of the urban variant [d] increased from about 30% in the local interview context to about 40% in the urban interview context.

5.2.1. The Influence of Age on Accommodation Patterns in the Use of (ð)

Similar to what was revealed for the overall data, age did not have a statistically significant influence on the use of (ð) variants across interview contexts. However, accommodation towards the urban interviewer occurred in the speech of most participants, apart from the oldest group, as evident from

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

Age Group	Interviewer	[ð]	%	[d]	%	[z]	%	Total
3–5	Local	17	40.5%	24	57.1%	0	0	42
	Urban	5	17.2%	22	75.9%	1	3.4%	29
6–8	Local	27	50.9%	26	49.1%	0	0	53
	Urban	9	25.7%	25	71.4%	1	2.9%	35
9–11	Local	47	78.3%	13	21.7%	0	0	60
	Urban	23	69.7%	10	30.3%	0	0	33
12–14	Local	22	71%	8	25.8%	1	3.2%	31
	Urban	13	61.9%	6	28.6%	2	9.5%	21
15–17	Local	54	100%	0	0	0	0	54
	Urban	35	92.1%	2	5.3%	1	2.6%	38

and Figure 7 below. This was most noticeable in the speech of 3–5 and 6–8-year-old speakers. For example, use of the local variant decreased from about 50% to about 25% in the speech of 6–8 year olds and from about 40% to about 17% in the speech of the 3–5 year olds.

5.2.2. The Influence of Gender on Accommodation Patterns in the Use of (ð)

Even though there were no significant differences in the speech of either male or female speakers across interview contexts,

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

Gender	Interviewer	[ð]	%	[d]	%	[z]	%	Total
Male	Local	101	82.8%	21	17.2%	0	0	122
	Urban	46	59.7%	30	39%	1	1.3%	77
Female	Local	66	55.9%	50	42.4%	1	0.8%	118
	Urban	39	49.4%	35	44.3%	4	5.1%	79

and Figure 8 below show that both male and female speakers used the local variant less in the urban interview context than they did in the local interview context, while they used the urban variants more frequently in the urban interview context, indicating an obvious trend of accommodation towards the urban interviewer. The difference in variant frequencies across interview contexts is more noticeable in the speech of boys since use of the local variant is relatively infrequent in the speech of female speakers even in the interview with the local interlocutor.

5.2.3. The Influence of Age and Gender on Accommodation Patterns in the Use of (ð)

A breakdown of the results by age and gender shows that accommodation towards the urban interviewer occurred in the speech of most participants, apart from male speakers in the oldest group (15–17) who used the local variant categorically in both interview contexts. This was most evident in the speech of 6–8-year-old males and females, as well as 3–5-year-old males. However, it is worth noting that the overall low frequency of the local variant in the speech of 3–15-year-old males might be compounded by developmental considerations, given the complexity of interdental fricatives in acquisition (e.g., Amayreh 2003). Results also revealed a surprising pattern in the speech of 9–11-year-old female speakers, whose use of the local variant increased in the urban interview context. This was likely due to small token numbers and relatively infrequent use of the local variant in their speech.

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

Male Speakers
Age Group	Interviewer	[ð]	%	[d]	%	[z]	%	Total
3–5	Local	3	21.4%	11	78.6%	0	0	14
	Urban	0	0%	16	94.1%	1	5.9%	17
6–8	Local	9	60%	5	40%	0	0	15
	Urban	4	30.8%	9	69.2%	0	0	13
9–11	Local	45	93.8%	3	6.3%	0	0	48
	Urban	18	81.8%	4	18.2%	0	0	22
12–14	Local	17	94.4%	1	5.6%	0	0	18
	Urban	9	90%	1	10%	0	0	10
15–17	Local	27	100%	0	0	0	0	27
	Urban	15	100%	0	0	0	0	15
Female Speakers
3–5	Local	14	50%	13	46.4%	0	0	28
	Urban	5	41.7%	6	50%	0	0	12
6–8	Local	18	47.4%	20	52.6%	0	0	38
	Urban	5	22.7%	16	72.7%	1	4.5%	22
9–11	Local	2	16.7%	10	83.3%	0	0	12
	Urban	5	45.5%	6	54.5%	0	0	11
12–14	Local	5	38.5%	7	53.8%	1	7.7%	13
	Urban	4	36.4%	5	45.55%	2	18.2%	11
15–17	Local	27	100%	0	0	0	0	27
	Urban	20	87%	2	8.7%	1	4.3%	23

and Figure 9 below exhibit the use of (ð) variants across interviews by age and gender.

5.3. Accommodation Patterns in the Use of (q)

As discussed above, (q) is primarily realized as [ɡ] in Bedouin dialects and as a glottal stop in Urban varieties. Given its status as a supralocal feature (Al-Wer and Herin 2011), it was hypothesized that use of the urban variant [Ɂ] would increase in the interview with the urban interlocutor while use of the local variant [ɡ] would, in turn, decrease in this context. A paired-samples t-test showed that the urban variant [Ɂ] was, in fact, used significantly more in the interview with the urban interlocutor than in the interview with the local interviewer at p = 0.006, while use of the local variant [ɡ] was significantly higher in the interview with the local interlocutor: p = 0.004. Little change occurred in using the standard variant since it is not associated with any spoken dialect in the context of this study, as can be seen in

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

Interviewer	[ɡ]	%	[Ɂ]	%	[q]	%	Total
Local	941	79.8%	94	8%	143	12.1%	1179
Urban	362	57.6%	178	28.3%	85	13.5%	628

and Figure 10 below.

5.3.1. The Influence of Age on Accommodation Patterns in the Use of (q)

Age played a key role in accommodation patterns in the use of (q). Differences in using the urban and local variants occurred in the speech of all age groups, apart from the oldest (15–17) who used the local variant overwhelmingly in both interview contexts. A significant difference in using the local variant [ɡ] appeared in the speech of the 9–11-year-old group who used the variant significantly less in the interview with the urban interlocutor at p = 0.019. Speakers in the 6–8-year-old group used the urban variant [Ɂ] significantly more in the interview with the urban interlocutor: p = 0.027.

Table 11. Distribution of (q) variants across interviews by age.

Age Group	Interviewer	[ɡ]	%	[Ɂ]	%	[q]	%	Total
3–5	Local	65	76.5%	12	14.1%	7	8.2%	85
	Urban	57	58.2%	29	29.6%	9	9.2%	98
6–8	Local	127	74.7%	22	12.9%	21	12.4%	170
	Urban	58	49.6%	42	35.9%	17	14.5%	117
9–11	Local	298	84.9%	12	3.4%	41	11.7%	351
	Urban	43	31.2%	65	47.1%	30	21.7%	138
12–14	Local	131	61.5%	47	22.1%	35	16.4%	213
	Urban	25	32.5%	37	48.1%	15	19%	77
15–17	Local	320	88.9%	1	0.03%	39	10.8%	360
	Urban	179	90.4%	5	2.5%	14	7.1%	198

and Figure 11 demonstrates the use of (q) variants across interview contexts by age. Realizations with [q] included categorial use of the variant in borrowings from SA, and variable use of the variant in lexical items, which may be realized with the urban or local variant such as [qamaɾ] ‘moon’ or [qɑleb] ‘heart’. The former was excluded completely from the analysis and the latter were excluded from the following figures to highlight the differences between local and urban realizations since accommodation patterns are most evident in the use of those variants.

5.3.2. The Influence of Gender on Accommodation Patterns in the Use of (q)

Accommodation in the use of (q) occurred in the speech of both male and female speakers, although differences in using the local and urban variants across interview contexts appeared to be more drastic in the speech of females, as

Table 12. Distribution of (q) variants across interviews by gender.

Gender	Interviewer	[ɡ]	%	[Ɂ]	%	[q]	%	Total
Male	Local	555	79.6%	57	8.2%	85	12.2%	697
	Urban	212	69.3%	51	16.7%	41	13.4%	306
Female	Local	386	80.1%	37	7.7%	58	12%	482
	Urban	150	46.6%	127	39.4%	44	13.7%	322

and Figure 12 indicates. Indeed, a paired-samples t test revealed that the difference in using the variants was significant in the speech of female speakers, but not significant in the speech of male speakers, indicating that girls accommodated their speech to the urban interviewer more than boys did. Girls used the local variant significantly less in the interview with the urban interlocutor: p = 0.024, and used the urban variant significantly more at p = 0.030. Male speakers used the urban variant more in the interview with the urban interlocutor, but the difference was not statistically significant p = 0.075.

5.3.3. The Influence of Age and Gender on Accommodation Patterns in the Use of (q)

A further breakdown by age and gender showed that various levels of accommodation towards the urban interlocutor occurred in the speech of most participants, as indicated by

Table 13. Distribution of (q) variants across interviews by age and gender.

Male Speakers
Age Group	Interviewer	[ɡ]	%	[Ɂ]	%	[q]	%	Total
3–5	Local	34	75.6%	9	20%	0	0	45
	Urban	38	74.5%	9	17.6%	4	7.9%	51
6–8	Local	32	64%	12	24%	6	12%	50
	Urban	43	57.3%	22	29.3%	10	13.3%	75
9–11	Local	254	89.4%	9	3.2%	21	7.4%	284
	Urban	29	59.5%	4	8.2%	16	32.7%	49
12–14	Local	108	65.9%	27	16.5%	29	17.7%	164
	Urban	13	38.2%	16	47.1%	5	14.7%	34
15–17	Local	127	82.5%	0	0	18	11.7%	154
	Urban	89	91.8	0	0	8	8.2%	97
Female Speakers
3–5	Local	31	77.5%	3	7.5%	6	15%	40
	Urban	19	40.4%	20	42.6%	8	17%	47
6–8	Local	95	79.2%	10	8.3%	15	12.5%	120
	Urban	15	35.7%	20	47.6%	7	16.6%	42
9–11	Local	44	65.67%	3	4.5%	20	29.8%	67
	Urban	14	15.7%	61	68.5%	14	15.7%	89
12–14	Local	23	46.9%	20	40.8	6	12.2%	49
	Urban	12	27.9%	21	48.8%	10	23.3%	43
15–17	Local	193	93.7%	1	0.5%	12	5.8%	206
	Urban	90	89.1%	5	5%	6	5.9%	101

and Figure 13 below. Such accommodation is most noticeable in the speech of female speakers in the 3–5, 6–8 and 9–11-year-old groups. Girls in these groups used the urban variant [Ɂ] more than the local variant [ɡ] in conversation with the urban interviewer. Use of the urban variant [Ɂ] with the urban interlocutor was also highest in the speech of female speakers in the 9–11-year-old group. Small differences appeared in the speech of 12–14-year-old females despite their accommodation to the urban interviewer because their use of the local variant was relatively low in both interview contexts. Modest accommodation occurred in the speech of the oldest group as both male and female speakers in the group used the local variant overwhelmingly in both interview contexts. Considerable accommodation also appeared in the speech of the 12–14-year-old male speakers, whose use of the local variant in conversation with the urban interviewer decreased by 27.7%, while their use of the urban variant increased by about 31%.

5.4. Accommodation Patterns in the Use of (-a)

As noted in the introduction above, the variable (-a), which represents the morphophonological ending of feminine nouns and adjectives in Arabic, is realized as [a] in Bedouin dialects, including the local dialect of the speech community, but is primarily raised to [e] in many urban and rural dialects of the Levant (Al-Wer 2007; Al-Wer et al. 2022). The raised urban realization is considered a characteristic feature of major urban centres in the levant such as Damascus (Lentin 2007), Amman (Al-Wer 2007), Beirut (Naïm 2007), and Jerusalem (Rosenhouse 2007). Raising of the variable, which is known as imala in Arabic (Al-Wer et al. 2022), is phonologically conditioned in urban Levantine varieties, whereby it is inhibited in the environment of back constants (pharyngeal, glottal, emphatics, and post-velars) which favour a low vowel (Versteegh 2001; Al-Wer 2007; Habib 2012). As such, while [e] is considered the default variant in Urban Levantine dialects, [a] is used in environments that inhibit raising, e.g., /wɑɾɑqɑ/ ‘paper’ would not exhibit final imala and would be realized as [waɾʔa] in Urban varieties. Raising may also be inhibited in the environment of /ɾ/, e.g., [le:ɾa] ‘Syrian currency’ (Durand 2011). Despite being viewed as a supralocal feature (Milroy et al. 1994) that is gaining ground in the region (Al-Wer et al. 2022), phonological conditioning that governs the raising of (-a) in Urban dialects is expected to hinder its acquisition and advancement in the community as it makes it a complex feature (Chambers 1992; Kerswill 1996).

Indeed, the local variant was used overwhelmingly in both interview contexts. However, some use of the urban variant [e] did occur in the interview with the urban interviewer, as can be noted from

Table 14. Distribution of (-a) variants across interviews.

Interviewer	[a]	%	[e]	%	Total
Local	1190	100%	0	0	1190
Urban	596	91.8%	53	8.2%	649

and Figure 14 below. The increase in using the urban variant [e] in the interview with the urban interlocutor, though numerically small, was found to be significant at p = 0.010.

5.4.1. The Influence of Age on Accommodation Patterns in the Use of (-a)

Some level of accommodation to the urban interviewer occurs in the speech of all age groups apart from the oldest, as use of the urban variant [e] only emerges in the interview with the urban speaker, albeit only slightly as preference for the local variant [a] remains overwhelming throughout. The most noticeable difference appears in the speech of the 9–11-year-old group, as illustrated in

Table 15. Distribution of (-a) variants across interviews by age.

Age Group	Interviewer	[a]	%	[e]	%	Total
3–5	Local	139	100%	0	0	139
	Urban	65	92.9%	5	7.1%	70
6–8	Local	162	100%	0	0	162
	Urban	94	89.5%	11	10.5%	105
9–11	Local	262	100%	0	0	262
	Urban	110	82.7%	23	17.3%	133
12–14	Local	244	100%	0	0	244
	Urban	84	86.6%	13	13.4%	97
15–17	Local	383	100%	0	0	383
	Urban	243	99.6%	1	0.4%	244

and Figure 15 below, but no significant differences in using the variants appear in the speech of any age group.

5.4.2. The Influence of Gender on Accommodation Patterns in the Use of (-a)

Table 16. Distribution of (-a) variants across interviews by gender.

Gender	Interviewer	[a]	%	[e]	%	Total
Male	Local	596	100%	0	0	596
	Urban	239	98.3%	4	1.7%	240
Female	Local	594	100%	0	0	594
	Urban	360	88%	49	12%	409

and Figure 16 below show that noticeable accommodation towards the urban interviewer only occurs in the speech of female speakers. Their use of the urban variant [e] increases significantly, at p = 0.021, in the interview with the urban interlocutor, although it remains quite low. No accommodation occurs in the speech of male speakers as they use the local variant [a] categorically with the local interviewer and near categorically with the urban interviewer.

5.4.3. The Influence of Age and Gender on Accommodation Patterns in the Use of (-a)

The two previous sections showed that in relation to age, accommodation mostly occurred in the speech of the 9–11 and 12–14-year-old groups, and in relation to gender, it was only noticeable in the speech of female speakers. This section analyses accommodation in relation to both age and gender and shows that despite no significant differences in using the variants across interview contexts in their speech, most accommodation to the urban interlocutor occurred in the speech of 9–11 and 12–14-year-old female speakers, as can be noted from

Table 17. Distribution of (-a) variants across interviews by age and gender.

Male Speakers
Age Group	Interviewer	[a]	%	[e]	%	Total
3–5	Local	65	100%	0	0	65
	Urban	25	89.3%	3	10.7%	28
6–8	Local	19	100%	0	0	19
	Urban	24	100%	0	0	24
9–11	Local	181	100%	0	0	181
	Urban	55	98.2%	1	1.8%	56
12–14	Local	144	100%	0	0	144
	Urban	29	100%	0	0	29
15–17	Local	187	100%	0	0	187
	Urban	103	100%	0	0	103
Female Speakers
3–5	Local	74	100%	0	0	74
	Urban	40	95.2%	2	4.8%	42
6–8	Local	143	100%	0	0	143
	Urban	70	86.4%	11	13.6%	81
9–11	Local	81	100%	0	0	81
	Urban	55	71.4%	22	28.6%	77
12–14	Local	100	100%	0	0	100
	Urban	55	80.9%	13	19.1%	68
15–17	Local	196	100%	0	0	196
	Urban	140	99.3%	1	0.7%	141

and Figure 17 below.

6. Discussion

The results above reveal obvious patterns of linguistic accommodation in the speech of Arabic-speaking children and adolescents experiencing dialect contact. Accommodation patterns in the speech of participants were impacted by a number of factors including age, gender, and the linguistic features being accommodated, as revealed by the results and as will be detailed in this discussion.

Although differences in variant distribution across the two interlocutors were not always significant, an obvious trend of convergence towards the urban interviewer occurred in the realization of all variables to varying degrees. Indeed, use of the local variant decreased from about 70% to about 54% in the case of (ð), from about 77% to about 54% in the case of (θ), from about 80% to about 58% in the case of (q), and from categorical use to about 92% in the case of (-a), while use of the urban variants increased. Convergence to an overtly prestigious variety, or what is referred to as upward convergence (Giles et al. 1991; Giles and Ogay 2007), is well attested in the literature, especially in situations of dialect contact (e.g., Miller 2005; Habib 2010). It can convey social and linguistic awareness on the part of speakers (Hinskens et al. 2005) and may express a desire for social mobility, belonging, and social approval (Giles and Ogay 2007). It might also indicate a positive perception of both the urban interlocutor and the urban variety as accommodative behaviour is shaped by speakers’ perceptions and attitudes towards their own variety and that of the interlocutor (Thomason and Kaufman 1988; Gasiorek and Giles 2012). Indeed, speakers tend to converge to those they like and respect or to those they may perceive as belonging to a socially desirable group in an attempt to be associated with them and their positive values (Giles et al. 1991; Giles 2008).

Convergence towards the urban interviewer occurred to varying degrees in the speech of both girls and boys, apart from male speakers in the oldest group. Such convergence was generally higher in the speech of female speakers and gender had an overall significant effect on accommodation patterns in the case of (q), (θ), and (-a). This is in line with previous research which suggests that female speakers are more likely to converge to their conversational partners (Namy et al. 2002; Giles and Ogay 2007; Lelong and Bailly 2011). On some occasions, convergence appeared to be quantitatively higher in the speech of boys than it was in the speech of girls. For example, use of the local variant of (ð) decreased by 23.1% across interview contexts in the speech of boys (from 82.8% with the local interlocutor to 59.4% with the urban interviewer), while it only decreased by 6.5% in the speech of girls (from 55.9% with the local interviewer to 49.4% with the urban interviewer). This was largely because girls generally used the local variants less than boys, even in the local interview context. Gender differences in accommodation appeared even in the youngest age cohort, though not consistently in all variables. For example, while the use of the urban variant of (q) increased from 7.5% with the local interviewer to 42.6% with the urban interviewer in the speech of 3–5-year-old girls, it remained relatively stable in the speech of boys in that age group. This result supports the assertion that gender differences in accommodation appear early on in the speech of children (Sheldon 1990; Robertson and Murachver 2003). This pattern persisted with all age groups, despite some notable exceptions. For example, convergence to the urban interlocutor appeared higher in the speech of 12–14-year-old boys, whose use of the urban variant of (q) increased by 30.6% across interview contexts (from 16.5% to 47.1%) compared to girls in the same age group whose use of the variant only increased by 8% (from 40.8% to 48.8%). Similarly, use of the local variant of (θ) decreased by 28.7% in their speech (from 83.9% to 55.2%) but only by 6.2% in the speech of girls (from 33.3% to 27.1%). As discussed above, this was mainly due to girls, especially in these age groups, using the local variants less than boys even in the local interview contexts, as is clear from variant frequencies in their speech.

A relatively similar pattern is reported by Van Hofwegen (2015), who examined accommodation patterns in 11–15-year-old African American children. Convergence in the speech of boys in her sample increased at the age of thirteen but decreased again at the age of fifteen, whereas girls’ accommodation patterns were consistent across all age points. Van Hofwegen (2015) also found that girls were more likely to accommodate to peers as well as unfamiliar interlocutors, which may explain the relatively modest convergence in the speech of females in the oldest group, given the urban interlocutor’s age as well as her position as a familiar figure in the community. The lack of convergence on the part of boys in the oldest group may also be due to the interlocutor’s age and gender as adolescent boys are more inclined to accommodate to peers in same-sex dyads (Tuten 2008; Van Hofwegen 2015). Research also shows that speakers who are intrinsically more variable are more likely to converge to their conversational partners (Lee et al. 2021), which may explain the diminished convergence in the speech of the oldest group compared to younger speakers in the sample. Indeed, the results show an overwhelming preference for local variants by 15–17-year-old participants, indicating little variability in their speech and leaving little room for convergence. On the other hand, the speech of participants younger than 15 is highly variable, leading to more convergence.

In addition to the quantitative results summarized above, accommodative behaviour in the speech of participants was manifested through various other strategies. For instance, on some occasions, convergence to the urban interviewer occurred as direct imitation, as indicated by the following example from the speech of a boy in the 3–5-year-old group:

(1)	Urban interviewer	hai	ɣazæ:le
		this	a deer
	This is a deer (Referring to a toy figurine)
	Child:	ɣazæ:le?
	A deer?

This interaction features the urban realization of the feminine marker (-a) in the pronunciation of the word [ɣazæ:le] ‘deer’. As indicated by the results, little variation occurred in the realization of this variable, which suggests that the child in this interaction was engaging in direct imitation. Predictably, overgeneralisations in an attempt to converge to the urban interviewer also occurred in the speech of some young children. For example, for ɡæto- ‘cake’ as borrowed and modified from French, a six-year-old boy used the word [ʔæto] in a clear overgeneralization of the urban [ʔ] for what he perceived as the local realization [ɡ] of (q). Overgeneralization of [ʔ] also occurred in the speech of a 9-year-old girl who used [maɁdu:s] for [magdu:s]2—a traditional Syrian breakfast staple—in a clear attempt to converge to the urban speaker. Various such ‘mistakes’, some of which result from not yet mastering the lexical or phonological conditioning of non-native variants (Kerswill 1995), occurred in the speech of participants as they aimed to converge to the urban interlocutor. For example, a 10-year-old girl, who had not yet mastered the lexical split in the urban realization of (ðˁ), used [zˁ] in the realization of (Ɂɑðˁɑːfɪɾ) ‘fingernails’, rather than [dˁ]. Interestingly, using the urban fricative realizations of interdental fricatives (i.e., [s] for (θ), [z] for (ð) and [dˁ] for (ðˁ)), whether erroneously or otherwise, was extremely sporadic in the data, which may suggest a general disfavouring of these variants as too urban (see Shetewi 2018 for more detailed insights). At the same time, this may indicate an earnest effort of convergence on the part of this young speaker, as well as a strong preference for urban realizations. Her realization is also an example of interdialectal or intermediate forms (Trudgill 1999), which were, expectedly, abundant in the data. In this instance, the speaker only modified her realization of (ðˁ), keeping the vocalic pattern of her native dialect rather than using the urban [Ɂɑdˁɑfi:ɾ]. Other such examples where the native vocalic structure is maintained and only one phonological feature is modified include: i) [Ɂalatli] ‘she told me’ for the urban [Ɂæ:litli] where only the realization of (q) is modified from the local [ɡ] to the urban [Ɂ], ii) [ʔanɑdˁdˁifha] ‘I clean it’ for the urban [bnɑdˁdˁifa] where only the realization of (ðˁ) is modified from the local [ðˁ] to the urban [dˁ]. Interdialectal forms where surface features are retained while the vocalic structure is changed also occurred in the data. For example, some speakers used the urban vocalic pattern in words like [ɡamaɾ] ‘moon’ and [baɡara] ‘cow’ in place of the traditional [ɡumaɾ] and [bɡara], but used the local realization of (q) rather than the urban glottal stop. In these examples, the local vocalic structure seems to be more indicative of ‘Bedouinesss’ than [ɡ] and is, therefore, abandoned. Indeed, the traditional vocalic pattern of these words never occurred as part of participants’ typical speech in the present study. One male speaker in the 15–17-year-old group jokingly used the word [baɡara] ‘cow’, which may indicate that such features are viewed as ‘outdated’ and are overtly stigmatized.

These examples suggest that (socio)linguistic competence has an impact on speakers’ ability to accommodate, as well as on the level of such accommodation (Pitts and Harwood 2015). Overgeneralization may, thus, occur in situations where speakers do not have full command of the sociolinguistic constraints of their interlocutor’s variety. This is especially true in the case of young children who do not have access to the full range of styles in language (Kerswill 1996). Giles et al. (1991) also note that accommodation can either be manifested on a large scale whereby a completely different mode of communication is adopted, or it can simply occur in a small aspect of speech but not in another. Likewise, speakers may converge to certain features while diverging on others. This is a unique characteristic of accommodation as it pays attention to both micro and macro communicative modes (Giles et al. 1991). Indeed, Trudgill (1986, 1999) also argues that convergence is about reducing differences rather than eliminating them and does not have to result in a complete change to one’s phonology. He proposes that such adjustments may result in intermediate forms, as evident from the examples above. What features may be subject to change depends on their social and linguistic constraints, as Kerswill (1995) explains that surface features that have sociolinguistic salience and are consciously recognised by speakers are the first to undergo change, whereas complex underlying features are harder to change. The examples above show that both surface and underlying features may be subject to change, which indicates that features that are regarded as more stigmatized are abandoned first, whether they are surface or structural features. Indeed, Hinskens et al. (2005) suggest that convergence may be achieved by either approximating linguistic forms of the other group or by avoiding one’s own marked features, and previous research shows that highly stigmatized features are, in fact, found to be particularly susceptible to change in contact situations (e.g., Miller 2005 on rural speakers in Cairo).

As observed above, in addition to being impacted by age and gender, accommodation patterns were also influenced by linguistic variables under examination. For example, convergence towards the urban interlocutor seems to be most drastic in the use of the (q). Use of the urban variant of (q) increases from 8.3% with the local interviewer to 48% in the interview with the urban interviewer in the speech of 6–8-year-old female speakers, whereas use of the local variant drops from 79% with the local interlocutor to 35% with the urban interlocutor. In the speech of 9–11-year-old girls, use of the urban variant of (q) rises from just 5% with the local interviewer to 69% with the urban interviewer and their use of the local variant drops from 66% in the interview with the local interlocutor to 16% in their speech with the urban speaker. This indicates that the variable is highly marked in the community as girls in these groups appear to be the most conscious of prestige features, as evidenced by their linguistic choices in general and in their speech to the urban interlocutor, in particular. Indeed, in conversations with adult members of the community, a few of them remarked that they made a conscious decision not to use [ɡ] with their children and to discourage them from using it as not to be ‘ridiculed’ when they eventually go to Damascus to study in the university.3 Conversely, little accommodation occurred in the realization of the feminine suffix (-a) despite the significant difference in variant frequencies across interview contexts. A closer look at accommodation patterns of (-a) revealed that convergence to the urban speaker was largely limited to girls in the 9–11 and 12–14-year-old groups, who were found to overwhelmingly favour urban variants throughout (see Shetewi 2018 for more detail). Limited accommodation, and indeed variation (cf. Al-Wer et al. 2022), in the realization of this variable may be due to the complex phonological conditioning involved in its urban realization, compared to the simple rule of no raising that applies in the local dialect. Kerswill (1995) explains that complex dialect features require early exposure for complete acquisition, and Miller (2005) observes that a high degree of difference between contact varieties complicates the process of accommodation. This feature may also carry what Trudgill (1986) refers to as “extra-strong salience”, and, as such, is not adopted by speakers in the community (Watt et al. 2010). Indeed, an adult female speaker from the community remarked that while she uses the urban variant of (q) in her speech, she would never use the urban variant of (-a). Evoking (-a) as a feature that she would not change was not prompted, so in response to a follow up question, she explained that using it would feel like going ‘too far’ and putting on a fake accent when her goal is to simply ‘tone down’ her accent and not sound so ‘Bedouin’ and she achieves that by abandoning [ɡ]. Patterns of accommodation in the realization of interdental fricatives were relatively variable, especially for female speakers between 6 and 14 years old, given the higher rates of adoption of the urban stop variants in their speech (e.g., in comparison to (q) and (-a)). Accommodation patterns in the use of interdental fricatives, in comparison to those in the use of (q), may also have been impacted by the retention of interdentals by the Urban interviewer. Still, obvious convergence to the urban interviewer appeared in realizations of these variables throughout (apart from speakers in the 15–17-year-old group). Interestingly, accommodation to, and indeed adoption of, urban variants in realizations of interdental fricatives was almost exclusively manifested through use of the urban stop variants (i.e., [t] for (θ) and [d] for (ð)), and not the urban fricative variants (i.e., [s] for (θ) and [z] for (ð)). As the split in urban realizations of interdental fricatives is primarily lexically conditioned, with the fricative variants being mainly used in lexical borrowing from SA (see Habib 2011a), it can be argued that speakers in this community have no need for those variants since they have access to the interdental fricatives in their own native repertoire (see Shetewi 2018 for further details). Indeed, the same adult speakers who expressed a positive attitude toward the urban variety and explicitly discouraged their children from using local variants (primarily [ɡ]) remarked that ‘we are able to pronounce interdental fricatives ‘correctly’ and have no ‘need’ for [s] or [z].

7. Summary and Conclusions

The results and discussion above give various important insights on the language development and linguistic practices of these children and adolescents, as well as shed some light on patterns of variation and change in the community.

For example, convergence to the urban speaker by children in the youngest age group, which is in line with previous research that finds accommodative behaviour to emerge early in children (e.g., Paugh 2005; Montanari 2009; Kaiser 2022), suggests a good level of social and linguistic awareness on the part of these children. Such convergence was manifested through direct imitation on some occasions, as noted in example one above. However, this does not discount the level of social and linguistic awareness on the part of children in this young group, but rather serves to show their ability to observe differences between their speech and that of the urban interlocutor (Hernández 2002; Paquette-Smith et al. 2019). Moreover, imitation, especially in phonetic convergence, is well-attested in the literature, even for adult speakers (Namy et al. 2002; Babel 2012; Nielsen 2011; Babel et al. 2014; Paquette-Smith et al. 2022). Indeed, children become aware of the social value of the linguistic features in their environment at an early age, which may, in turn, impact their language use (Cornips 2020). On the other hand, previous research also shows that children’s ability to adapt their speech to context/interlocutor is not dependent on their knowledge of the social value of linguistic features (Chevrot et al. 2000). So, while the results show young children’s ability to accommodate and may suggest a positive evaluation of the urban interlocutor and dialect (Giles et al. 1991), their patterns of accommodation may simply be influenced by their input, which in this speech community is mostly female-oriented (see Shetewi 2018 for more detail). As such, convergence in their speech might denote conformity to what is perceived as ‘female’ speech patterns, which is a conclusion that is supported by the overall preference for urban variants in the speech of this group.

Such preference for urban variants persists, and even increases, in the speech of girls between 6 and 14, while an opposite trend emerges for boys in the same age groups. This results in some drastic differences in rates of accommodation between girls and boys in this age range, as detailed in the previous section. This is because, despite their overall preference for local variants, boys in this age range exhibit considerable convergence towards the urban interviewer. The higher overall use of the urban variants by girls in this age range may indicate long-term accommodation, given that they adopted such variants in their own speech and are using them outside of interactions with urban interlocutors (Auer and Hinskens 2005, p. 336). Likewise, higher proportions of urban features in the speech of the youngest group, presumably reflective of their primarily ‘female-oriented’ input, may reflect long-term accommodation in the speech of their primary caregivers (see Shetewi 2018 for more details). Indeed, given the model of dialect contact in this speech community, which is in the form of geographical diffusion as explained in Section 3 above, children’s variable input, which is most clearly reflected in the speech of the least mobile participants, is indicative of language change in the community. While it may not be reliably concluded, based on the results from this dataset, that such language change is a result of interpersonal accommodation, some aspects of dialect levelling are ‘foreshadowed’ (Auer and Hinskens 2005, p. 347) in some of the accommodation patterns manifested in this study.

Conversely, accommodation patterns in the 15–17-year-old group, with slightly decreased convergence in the speech of girls and overwhelming maintenance in the speech of boys, may also indicate a conscious effort to conserve group identity (Bourhis 1984; Giles and Ogay 2007), especially given the prominence of identity practices in the lives of adolescents (e.g., Van Hofwegen 2015; Tuten 2008). They indicate a positive attitude towards the local variety and community. In the context of this study, such an attitude may have been enhanced by the fact that interviews were carried out in participants’ houses and, as such, these speakers were operating within their physical and emotional space, while the urban interviewer was viewed as the outsider in the situation. Moreover, convergence to an overtly prestigious norm is argued to be an attempt at membership of a socially attractive group (Giles and Ogay 2007), which is a sentiment that is not expressed by this age group, as evident by their overwhelming preference for the local variants throughout the data (Shetewi 2018, 2023). In fact, adolescents are found to be rebellious rather than seeking approval and integration, especially when interacting with adults (Labov 2001; Eckert 2017). They use language to construct independent social identities and express belonging to peer groups, rather than identify with adults (Eckert 2004, pp. 112–13), and such strong identity practices often manifest in divergence from adult speech norms (Garrett and Williams 2005). In dialect contact settings, such focus on constructing independent identities coupled with a general tendency towards the vernacular may manifest in adolescents preserving their own dialect features. This is attested in situations of contact involving minority and dominant varieties where a local/ethnic orientation is an especially strong index in peer group affiliation (Rampton 1995; Van Hofwegen 2015) and may, by extension, apply in situations involving regional or national identities, such as in the case of the present study. Indeed, speakers in this group seem to be using language to index a Palestinian identity. They seem to view linguistic variation in their environment as not simply urban vs. Bedouin but rather Syrian vs. Palestinian (Shetewi 2018, 2023). As such, it is not surprising to find maintenance and even divergence patterns in the speech of adolescent boys, especially. Indeed, one of the male speakers in this group seemed to diverge his speech even further from the urban interlocutor after she commented that it is ħɑrɑ:m ‘forbidden in Islam’ for him to become a hair stylist. This divergence was expressed by affricating (k) into [tʃ] in the second singular feminine suffix, a feature that did not occur in his speech with the local interlocutor and one that may be viewed as ‘outdated’ and ‘too traditional’. Indeed, this feature only occurred twice in the data set, with the other occurrence being in the speech of a 17-year-old girl who laughed as she used it and commented embarrassingly that people laugh when she ‘talks like this’, indicating that it is viewed negatively by some people. The speaker’s divergence on this occasion appeared to signal a conscious effort to further distance himself from the interviewer in response to a negative comment she made about his career goals (Giles et al. 1991; Zhang and Giles 2018). Convergence in the speech of boys younger than 15 also implies that a strong sense of local identity is more pronounced in adolescents in the community regardless of age.

The above observations presume that children’s and adolescents’ accommodative behaviour in this study is socially motivated. This is supported by various indications in the data, including divergent behaviour in the speech of 15–17-year-old speakers. More importantly, low rates of accommodation in the realization of (-a), which is consistently realized as [e] by the urban interviewer, compared to the high rates of accommodation in the realization of (θ) and (ð), which are overwhelmingly retained in her speech, suggest that convergent behaviour in the speech of participants is not always motivated by the objective speech patterns of the interlocutor but rather by the expectations of how she should sound, based on certain stereotypical features. Wade (2022) refers to this as ‘expectation-driven’ accommodation as opposed to ‘input-driven’ accommodation and notes that while the latter might occur as an automatic response, the former is largely based on social motivations. This is well-attested under the framework of Communication Accommodation Theory, whereby accommodative acts are often based on underlying beliefs and attitudes, which do not always match objective reality (Thakerar et al. 1982; Dragojevic et al. 2016). Listeners use their conversational partners’ speech norms, especially phonological features (Coupland 1985), to form their own social perceptions of them, which, in turn, would impact their own accommodation patterns (Gasiorek and Giles 2012; Garrett 2010; Auer and Hinskens 2005). This also echoes Bell’s (1984, 2001) referee design, whereby speakers may adapt their speech beyond their conversational partners based on their own perceptions and associations of the interaction. According to Bell (1984, p. 185), the shift in speakers’ language use in such cases is ‘initiative’ rather than ‘responsive’, and often involves hyperconvergence beyond the interlocutors’ own speech patterns. This suggests that children’s convergent behaviour in this current study is primarily socially motivated. The overgeneralizations and speech ‘mistakes’ noted above support this conclusion and show that, although not fully competent in the urban variety they attempt to emulate, children in the community use the linguistic resources available to them in a strategic and socially appropriate manner. This study examined accommodation patterns across a wide range of ages encompassing children as young as three, at the early stages of structured variation, all the way up to the last year of secondary school, which marks the threshold of sustained mobility that follows it. Further research that examines accommodation patterns in the speech of adult members of the community, who are generally more mobile and experience more face-to-face contact with speakers of other varieties, may uncover interesting insights on the linguistic practices and language attitudes that may result from a different mode of contact. Additionally, the mobility caused by the unfortunate events in Syria offers further opportunities for examining such themes in the speech of children and adolescents who now reside in nearby localities outside of the speech community. Diverse modes of contact, in addition to varying considerations of identity, may have different implications on their language use but will, equally, offer important insights on their linguistic development.