Degemination in Emirati Pidgin Arabic: A Sociolinguistic Perspective

Abstract

This study examines the production of geminates in Emirati Pidgin Arabic (EPA) spoken by blue-collar workers in the United Arab Emirates. A simple naming test was designed to test the production of geminates to determine whether the EPA speakers would produce a geminated or degeminated phoneme. Following that, a semi-structured interview was conducted with a subset of the study cohort to obtain the participants’ own explanation of where they degeminated the consonants. Our findings suggest that the exercising of this choice functions as a sociolinguistic strategy, akin to the one observed by Labov in his study of Martha’s Vineyard. In particular, our findings show that speakers of EPA are inclined to degeminate consonantal geminates to establish themselves as members of a particular social group. The reasons for wanting to achieve this aim were given as follows: to claim privileges only available to members of this group (such as employment); and to distinguish themselves from the dominant cultural group. The study concludes that degemination in EPA has developed into a sociolinguistic solidarity marker.

1. Introduction

The movement of people across the world for employment, trade or other reasons is known to give rise to what have been termed ‘contact languages’ where there is a pragmatic need for people from diverse language backgrounds to communicate with one another. Pidgins are languages that

“develop out of a need for communication among people who do not share the same language… most of the forms in the lexicon of the new language come from one of the languages in the contact situation, called the lexifier or sometimes the superstrate-usually the language of the group in control of the area where contact occurs.”

(Siegel 2008, p. 1)

A unifying characteristic of pidgins is that they derive from a dominant language and loosely retain elements of the dominant language’s phonology, structure, and lexicon (for Arabic-based pidgins, see Gomaa 2007).

Gulf countries currently attract significant cohorts of unskilled or semi-skilled migrant workers, who reside in the region on a temporary, and sometimes long-term, basis. These workers come primarily from Asian and South Asian countries including India, Pakistan, and the Philippines, and speak a number of languages such as Hindi, Pashto and Tagalog. The need for a lingua franca amongst these linguistically diverse communities has given rise to several pidgin languages. In the Arabian Gulf area, these pidgins are collectively known as Gulf Pidgin Arabic, and can be further broken down into distinct varieties. Emirati Pidgin Arabic (henceforth EPA), spoken in the United Arab Emirates (henceforth UAE), is one of these pidgins.

One of the phonemic features of EPA which distinguishes it from the local form of Arabic, Emirati Arabic (henceforth EA), is the divergent pronunciation of geminated consonants. Gemination refers to the ‘doubling’ or prolongation of consonantal phonemes. Geminated consonants form part of the phonemic repertoire of all Arabic varieties including EA, such that some consonantal phonemes can have both singleton and geminate realizations, and this functions as a contrastive feature in EA, e.g., mara ‘woman’ vs. marra ‘one time’. EPA is observed to have a strong preference for degemination of geminates in words of EA origin; for example, when pronouncing the word for ‘air-conditioner’ mukayyif, EPA speakers substitute it with mukayif.

In this study we examine degemination in EPA and seek to identify the factors for its preference in EPA, especially the influence of the speakers’ first languages in terms of being a geminating vs. a non-geminating language. While the speakers’ native languages must play an undeniable role in degemination in EPA, we claim that degemination also has a social function and is thus in part a conscious and deliberate strategy assumed by EPA speakers. We put forward the hypothesis that degemination in EPA may be functioning as a sociolinguistic marker of solidarity within the migrant worker community where it is spoken. Based on the speakers’ own commentaries, we also propose some initial suggestions as to why EPA speakers resort to this strategy to signal that they are ‘opting in’ to the EPA community and to distance themselves from the hegemonic EA speaking community.

The organization of this paper is as follows: in Section 2, we provide a general background of the study including key concepts, the theoretical framework adopted in this study, and previous studies on pidgins spoken in the Gulf area. In Section 3, the methods adopted to collect and analyze the data are discussed. Section 4 presents the results and discussion. In Section 5, we provide our conclusions.

2. General Background

2.1. Key Concepts

2.1.1. EPA

Pidgins develop as a pragmatic means of communication between people sharing no common language (Holmes 2013), typically in a trading or employment context. This simplified form is spoken only as a second language, and is derived from a dominant language, the ‘lexifier language’ (Siegel 1997). It features reduced complexity compared to that language. In general, a pidgin language is different from its lexifier in most linguistic levels, namely, phonological, morphological, syntactic, and lexical. It also has the tendency to be a simplified version of its lexifier in most linguistic aspects (Aljutaily 2018). For instance, pidgin phonology is significantly shaped by language universals (Holm 1989). That is, one of the most common features in the phonology of pidgins is that the pidgins, compared to their lexifiers, have a reduced inventory of phonemes. Such reduction is mainly ascribable to the highly marked sounds, i.e., those that are less common in the world’s languages and more challenging to pronounce (Aljutaily 2018), which are replaced by the closest equivalents in the substrate languages (Aljutaily 2018).

Gulf Pidgin (GP)1 is spoken in the Arabian Gulf; it is the general name given to all Arabic-based pidgins in the Arabian Gulf, among which is EPA, spoken specifically in the UAE. Expatriate workers in the area come from a variety of language backgrounds including Farsi, Panjabi, Malayalam, Urdu, Hindi, Bengali, Thai, Tagalog, Indonesian, Nepalese, and Tamil (Bakir 2010); thus, it is expected that pidgin forms of Arabic have developed among these workers who have little to no knowledge of Arabic. The ongoing flow of expatriates to the Arabian Gulf countries (Bakir 2010) starting from 1973, and thus the emergence of a sizeable community that is linguistically heterogeneous (Holm 1988, p. 5) has helped in sustaining GP in the area.

The consonant inventory of Gulf Pidgin (GP), including EPA, is significantly limited when compared to its lexifier language, EA, whereby the marked consonants of EA are either substituted or lost (Salem 2013). The GP consonant inventory contains all but the following Gulf Arabic (GA) consonants: uvular stop/q/, emphatic consonants/tˤ, dˤ, sˤ/, uvular fricatives/x, ʁ/, dental fricatives/θ, ð/, and pharyngeal fricatives/ħ, ʕ/ (Aljutaily 2018, p. 30). These marked (emphatic) consonants do not appear in the speech of GP speakers and hence, they tend to be substituted with their closest counterparts, i.e., usually their non-emphatic counterparts. This may be true of/tˤ, dˤ, sˤ/, but what about the others, e.g., /θ, ð/? The vowel system of GP is also reduced in comparison to its lexifier (GA); it consists of five vowels/i, e, a, u, o/compared to the eight vowels of GA/i, i:, e, a, a:, o, u, u:/. GP also lacks the vowel distinction that takes place in GA, in particular, speakers of GP are affected by their L1s and this is evident in their vowel length which is neutralized by GP speakers. Hence, the lack of phonemic length in the speakers’ L1 yields reduced vowel inventories in GP (Aljutaily 2018). This suggests that taking the substrate languages of pidgin speakers into account helps in exploring L1 influence when investigating pidgins. The current study deals with the variant of Pidgin Arabic spoken by migrant workers in the UAE, Emirati Pidgin Arabic. It is based on the Emirati variety of Arabic.

2.1.2. Emirati Arabic

EA, one of the Gulf dialects of Arabic and the lexifier of EAP, is spoken in the UAE. It has numerous phonological, morphological, and semantic features that distinguish it from other dialects of Arabic. (See Hassan 2017).

2.1.3. Geminates in Arabic

In Arabic, geminates can occur word medially or word finally (Davis and Ragheb 2014) as the following two examples in

Table 1. Examples of geminates in medial and final positions in Arabic.

Position	Example
Medial position	/tˁab’baax/‘cook’
Final position	/sadd/‘dam’

show.

Note also that syllables with geminated phonemes tend to attract word stress, especially in the medial position2. Degemination therefore also has an impact on word stress patterns. This feature was also given attention to in our study.

2.2. Theoretical Framework: Communication Accommodation Theory (CAT)

Communication Accommodation Theory (CAT) proposes that interlocutors adjust their speech to create or maintain, or to decrease, social distance and to signal their attitude towards others (Giles and Ogay 2007). It is a useful framework to explain the dynamics of language change, particularly in the context of contact languages.

Convergence strategies are those that individuals use to adjust their communication to be more like that of their interlocutor (Giles 1973; Giles and Ogay 2007). These strategies are rewarding in the sense that they can lower uncertainty and interpersonal anxiety and foster mutual understanding among interlocutors (see Gudykunst 1995). However, this may be at the expense of individual personal or social identity (see The Social Identity Theory by Tajfel and Turner 1986). Divergence, on the other hand, refers to the strategy individuals use to emphasize their distinctiveness based on group membership, to preserve shared social identity and to enhance ‘in-group pride’ and ‘feelings of self-worth’ (Giles and Ogay 2007).

One of the best-known studies of a divergence strategy is William Labov’s classic investigation of language shift on Martha’s Vineyard (Labov 1972). In his study, Labov (1972) investigated the production of two diphthongs, (/aʊ/and/aɪ/) as produced by residents of the area. Labov noticed that the speech of the local population tended to deviate from the ‘standard’ one in the pronunciation of the two diphthongs—instead of the standard pronunciation, the speakers adopted pronunciation with various degrees of centralization (closer to [əʊ] and [əɪ] respectively). Labov’s sociolinguistic investigations concluded that the variant of pronunciation that he noticed was a deliberate strategy of speakers to identify themselves as locals and to differentiate themselves from visiting holidaymakers, whose mass arrival in the summer months was sometimes resented by the locals.

For context, the next section presents some previous studies on pidgin in the Gulf area countries.

2.3. Previous Studies on Pidgins Spoken in the Gulf Area

Tracing back the history of pidgins in the Arabia Gulf demonstrates that the oil boom in October 1973 marked the beginning of the emergence of Gulf Pidgin (GP). The Arab Gulf countries witnessed radical demographic, social and political changes in a very short period (Albaqawi 2016). These changes led to the rapid increase in the demand for foreign labor. The reason for this high demand was that the Gulf national workforces were not ready to execute huge projects as they were too small and lacked the required skills. Because of this, during the oil decade bonanza between 1973 and 1982, the numbers of non-indigenous workers in the Gulf states remarkably increased to reach 4.4 million in 1985. According to Albakrawi (2012), more than 7 million workers from Asian countries work in Saudi Arabia. The native language or Arabic variety spoken in the Arabia Gulf is GA, which is a variety of colloquial Arabic spoken by the native people of the Gulf region (for more information see Avram 2016). These countries constitute a multilingual setting; migrant workers from different countries, who belong to different linguistic backgrounds and do not speak Arabic, migrate to Gulf countries to look for job opportunities. As a result, they come into contact with GA speakers and speakers of other Arabic varieties. Since Arabic speaking locals and expats, on the one hand, and non-Arabic speaking expats, on the other, do not share a language and there is an urgent need to communicate (Almoaily 2012, p. 1), a simplified form of Arabic has emerged due to the contact between the above groups. This form of Arabic is known as GP; it is a reduced language system spoken in the Gulf countries in the Middle East by foreign workers and native speakers of Arabic.

Pidgins, in general, were categorized by Bakker (1995, pp. 27–28) in terms of the social situation in which they are spoken: (1) trade pidgins, (2) maritime pidgins, (3) work force pidgins, and (4) interethnic contact languages. However, according to Avram (2012), all of the above varieties can be classified under the category ‘workforce pidgins’. Another classification was suggested by Sebba (1997, pp. 26–33); his typology was based on the social context of the language’s origins including: (1) seafaring and trade pidgins, (2) military and police pidgins, (3) mine and construction pidgins, (4) plantation pidgins, (5) tourist pidgins, (6) immigrants’ pidgins, and (7) urban contact vernaculars. Avram (2012) explains that the above categories can be all assigned under the category ‘immigrants’ pidgins’. Thus, pidgins in general and GP in particular qualify for the group ‘immigrant workforce pidgin’.

Even though major advances in the area of pidgins and creoles have been accomplished over the past 50 years, more research on pidgins based on non-Indo European languages is still needed (Almoaily 2012). Almoaily (2012) suggested that there is still a lack of documentation and examination of non-Indo European language-based contact languages. The examination of pidgins in general has started to gain momentum relatively recently. Similarly, few studies have explored the linguistic characteristics and the sociolinguistic situation of GP except for some descriptive papers (e.g., Alshammari 2010; Almoaily 2012; Avram 2014, 2015, 2017, 2020, among others). For instance, Avram (2017, 2020) provides different features of GP in Gulf countries, one of which is the UAE. In another study, Avram (2012) explored the functions of fi ‘at, exist’ in four Arabic lexified pidgins, i.e., Saudi Pidgin Arabic, Pidgin Madam (spoken in Lebanon by Sri Lankan housekeeping female workers), Omani Pidgin Arabic, and Qatari Pidgin Arabic. A corpus of answers to questionnaires, transcripts of interviews, translations of test sentences and various online sources was collected to be analyzed by the researcher. The analysis showed that fi functions as an existential copula in possessive-have constructions, as a locative copula and as a predicative copula. This use of fi in the target pidginized Arabic varieties was ascribed to the role of grammaticalization and the influence of the substrate languages.

In another recent study, Al-Azraqi (2020) conducted a study on a pidgin mainly spoken by Asian immigrants in Abha, Saudi Arabia to investigate the multifunctionality of three grammatical categories, namely, pronouns, predication, and definiteness. Even though other studies conducted on GP have described this variety in terms of multifunctionality (e.g., Bakir 2014), Al-Azraqi’s (2020) study puts more focus on the different pathways that manifest this multifunctionality. The study provides a description of three types of multifunctionality in terms of generalization, refunctionalization, and neutralization. The data was collected via audio-visual interviews with a random sample consisting of 24 Asian (male and female) participants whose ages ranged between 26 and 45 years of age. They occupied different jobs and their length of stay in Saudi Arabia was between four and nine years. The analysis shows that fi has several functions: it is used as an existential marker with the meaning ‘there is/are’, a preposition, and a generalized predicate marker before nominal, adjectival and verbal predicates. In addition, hada ‘this’ was shown to be refunctionalized to mark definiteness and to be used as a deictic marker to act as a demonstrative. Finally, the pronominal system of the investigated pidgin exhibits simplification from ten to five pronouns only. These studies may demonstrate that that the main characteristics of GP are multifunctionality and simplification.

Previous studies on GA and GP show that they are two distinct varieties of Arabic exhibiting phonological, morphological, syntactic, and semantic differences. Yet, in addition to the indigenous colloquial GA, these non-Arabic speaking expats come in contact with other languages, language varieties, and registers while this contact plays a role in determining the features of GP. Consequently, various pidginized forms of Arabic have emerged across the Arab world and mainly in the Arabian Gulf (Albaqawi 2016). Bizri (2014) explains that it is still unknown as to why Asian expatriates who work under similar conditions in the Gulf countries apparently develop dissimilar pidgins, e.g., Saudi Pidgin Arabic, Emirati Pidgin Arabic, and Omani Pidgin Arabic among others. Thus, there is a need to explore the differences between these pidgins to obtain a better understanding of the similarities and differences between them and to determine why there are differences between them. This study aims to contribute to the existing literature on GP by examining the production of geminates in EPA which is a subvariety of GP.

The present study takes a novel approach; we investigate the factors that affect one of the distinctive phonemic features of EPA, namely, the degemination of EA geminated phonemes. The goals of the study are to address the following two questions:

(a)

To what extent do speakers of EPA use geminates?

(b)

What factors affect speakers’ degemination of geminates in EPA?

The next section describes the methodology used in the current study.

3. Methodology

3.1. The Participants

The participants taking part in the current study were 35 workers of the following nationalities: Indian (n = 14), Pakistani (n = 13), and Filipinos (n = 8), speaking Malayalam (n = 11) or, Hindi/Urdu (n = 10), Pashto (n = 7), and Tagalog (n = 7), respectively (see Appendix A for more details). While Hindi/Urdu, Malayalam, and Pashto have geminate consonants (see Khan 2002; Hussain 2012; Aarti and Kopparapu 2018; Ullah et al. 2020), Tagalog does not (see Reid 2009)3. We will consider whether the existence of geminates in the speakers’ L1s could have an effect on their geminate vs. degeminate realizations of the study items.

The reason behind choosing these nationalities was that people from these nationalities constitute the biggest number of expatriates coming to the UAE for work; the Indian expatriate community alone constitutes the largest community in the UAE for oil and non-oil trade (Alnamer and Alnamer 2018). As for the participants’ residency in the UAE, 14 of them (40%) have been in the UAE for 5–8 years while the rest have been residents for more than 20 years.

The UAE provides many job opportunities, be they domestic ones or ones in production, construction, repair, transportation, food processing, or manufacture. Given that the local community members mostly interact on a daily basis with expatriates working in the above services, the participants were chosen accordingly: 10 participants were barbers, 9 were working in goods and food delivery, 6 were taxi drivers, 5 were tailors, and 5 were mechanics.

It is important to mention that these participants were not holders of any academic degree, or high school diploma, and bearing this fact in mind along with the nature of their jobs, they could be aptly referred to as ‘blue collars’ in the current study.

Of these 35 participants, 26 (i.e., 74%) were practicing Muslims with a knowledge of Koranic Arabic. In the course of their religious practice, they regularly read the Koran and recited prayers aloud in Arabic. This is significant for reasons which will follow.

3.2. Data Collection

The current study adopts a mixed-method approach. Both quantitative and qualitative data were collected to benefit from the strength of each of these approaches whilst mitigating their respective weaknesses (Denscombe 2014; Creswell and Creswell 2017). A naming test was used to collect quantitative data, and semi-structured interviews generated qualitative information. The next section discusses these two data collection methods in further detail.

3.2.1. The Naming Test

A naming test was designed to elicit spontaneous verbal responses. Thirty participants (Malayalam (n = 10) or, Hindi/Urdu (n = 8), Pashto (n = 6), and Tagalog (n = 6)) took the test, which consisted of images of ten objects. All target responses were one-word concrete nouns (the test pictures are included in Appendix B). When pronounced in EA, each of these objects’ names has a stressed syllable with a geminated consonantal phoneme.

Table 2. The names of the objects used in the study test and their meaning.

Words	Meaning
/ʃaˈɣɣaala/shaghghaala	maid
/ɣaˈssaala/ghassaala	washing machine
/xaˈjjaatˤ/khayyaat	tailor
/tˤaˈjjaara/tayyaara	plane
/saˈjjaara/sayyaara	car
/daˈwwaar/dawwaar	roundabout
/raˈjjaal/rayyaal	man
/θaˈllaadʒa/thallaaja	fridge
/mukaˈjjif/mukayyif	air conditioner
/baˈqqaala/baqqala	grocery

below shows the names of the objects chosen for the study test and their meaning in English.

To supplement the data obtained through the naming test, five of the study participants (Malayalam (n = 1) or, Hindi/Urdu (n = 2), Pashto (n = 1), and Tagalog (n = 1)) were asked instead to name the same objects when these were observable in the interlocutor’s immediate environment. The same words/names were also produced in the course of conversation with the research team in order to avoid any self-conscious attempts to ‘correctly’ pronounce the words, which we judged may have been more likely to occur in the picture naming ‘test’ situation. However, such tokens were not included in the analysis of the current paper. The conversation was recorded for later analysis. To avoid the Observer’s Paradox4, the participants were not informed of the exact purpose of the study; they were just told that the study investigates a linguistic phenomenon. Consent forms were given by the participants to allow the recordings of their interviews. The data obtained through these recordings were carefully analyzed. The geminate vs. degeminate realizations were rated impressionistically by five raters, i.e., the three researchers and two other linguists working in two different UAE universities.

3.2.2. The Semi-Structured Interview

Data was also gathered through semi-structured interviews with 20 members of the study cohort. The semi-structured interview is a data collection method frequently used to gather qualitative, rather than quantitative, data (see Da Silva 2011; Al-Zubeiry 2015; Holloway and Galvin 2016; Catedral 2018; Altakhaineh et al. 2019). It provides the researcher with additional flexibility in the data collection process (Potter and Hepburn 2005; Bernard 2017; Roulston and Choi 2018). In this part of the study, interviewees were encouraged to discuss the responses they had given in the naming test (Hitchcock and Hughes 1995).

4. Results and Discussion

Results of the research indicate that, overall, when speaking EPA, the subjects were consistent in degeminating EA geminate consonants. A number of reasons were provided by the participants for this aspect of their pronunciation. These reasons are set out below. Overall, however, it can be proposed that this choice has the characteristics of a sociolinguistic divergence tactic akin to the one described by Labov. This is suggested on the grounds that the participants’ speech diverged from that of local Emiratis who use geminates; the participants degeminated the EA consonants to present themselves as true speakers of EPA. We propose that EPA speakers wish to emphasize identification with one linguistic community whilst, at the same time distancing themselves from another.

Table 3. Instances of gemination and degemination in the words of the study test by 30 participants.

Emirati Arabic Words	Consonant EA Gemination Retained	Consonant Degeminated
/ʃaˈɣɣaala/‘maid’	4 (13%)	26 (87%)
/ɣaˈssaala/‘washing machine’	3 (10%)	27 (90%)
/xaˈjjaatˤ/‘tailor’	2 (7%)	28 (93%)
/tˤaˈjjaara/‘plane’	3 (10%)	27 (90%)
/sajjaara/‘car’	3 (10%)	27 (90%)
/daˈwwaar/‘roundabout’	4 (13%)	26 (87%)
/raˈjjaal/‘man’	4 (13%)	26 (87%)
/θaˈllaadʒa/‘Fridge’	1 (3%)	29 (97%)
/mukaˈjjif/‘air conditioner’	7 (23%)	23 (77%)
/baˈqqaala/‘grocery’	9 (30%)	21 (70%)
Total	40 (13%)	260 (87%)

shows that geminate EA consonants were produced as singletons in 87% of instances. It should be noted that the participants with Hindi/Urdu or Malayalam as their first language, which all have geminate consonants, exhibited higher retention of geminate consonants, in the word baqqaala ‘grocery’, which happens to be borrowed into Urdu, as baqqal means ‘grocer’. The higher percentage of degemination, however, overshadows the retention of geminate consonants as indicated above. Despite the limited scope of the study in terms of sample size, we conclude that speakers of EPA show a clear tendency to degeminate consonants that are geminated in EA5.

Table 4. Consonant EA gemination vs. degemination based on speakers’ L1.

Speakers’ L1	Consonant EA Geminate Retained	Consonant Degeminated
Malayalam (n = 10)	17	83
Hindi/Urdu (n = 8)	14	66
Pashto (n = 6)	9	51
Tagalog (n = 6)	0	60
Total	40	260

below shows consonant EA gemination vs. degemination based on the speakers’ L1.

Table 4 shows that even though Hindi/Urdu, Malayalam, and Pashto have geminate consonants, degemination was prevalent in the reported data (40 geminates vs. 260 degeminates). The only difference is that those speakers whose L1 is Tagalog did not produce any geminates, while speakers whose L1s are Hindi/Urdu, Malayalam, and Pashto produced a small number of geminates compared to the degeminates. This may show that the influence of the speakers’ L1 is not an important contributing factor to the geminate vs. degeminated consonant realizations in the current study.

We also found during both the naming test and in the recorded conversation, a shift in syllable stress occurred from a penultimate or final syllable to the first one [i.e., leftward stress shift] in EPA. This unexpected shift was further checked by having 10 native EA speakers to listen to the respective excerpts and to judge the position of stress as well as whether the consonant is short or long.

Table 5. The shift of the stress into the first syllable in the case of degemination.

EA	EPA
[ʃaˈɣɣaala] ‘maid’	[ˈʃaɣaala]
[xaˈjjaatˤ] ‘tailor’	[ˈxajaat]

below illustrates this concomitant shift.

As for EA words that have geminated consonants in a final position and that occur before words with a definite article in EA, such as:

dagg	il-baab
knock.3SG.M	DEF-door
‘He knocked on the door’.

The participants degeminated the geminate consonants and did not pronounce the definite article. Hence, the previous example was produced as/dag baab/in EPA.

Degemination was prominent despite the fact that the majority of native languages spoken by the participants in the present study have geminate consonants. This may suggest that both groups of speakers whose first languages are Hindi/Urdu, Malayalam, and Pashto, on the one hand, and Tagalog, on the other opted not to use geminates in EPA. A plausible argument can therefore be put forward that EPA degemination is attributable to more than L1-influence/shift-induced interference/imperfect learning. Recall that the two classes of participants, namely those whose mother tongue has phonemic geminates and those speakers whose mother tongue did not contain phonemic geminates both degeminate the target words almost equally. This reduction could be ascribed to the desire to establish a certain social identity in their community of practice. This is also reported by the speakers themselves; according to them, this feature of EPA, namely degemination, is a clear signal to others of their membership of the EPA community. The fact that there still are retained geminated consonants for some speakers can be attributed to the fact that most of these participants, even the speakers of non-geminating L1s, reported that they did geminate when reciting the Koran. This illustrates that a measure of control over this aspect of pronunciation is still available to speakers of EPA at some level.

The detailed results of the interviews can be summarized as follows: the analysis of the interview data yielded significant additional information since the participants were asked to explain their language choice (i.e., their use of geminates vs. degeminates). Interestingly, their responses overlapped in three main areas. First, some of the participants declared that they preferred degemination due to its ease of articulation (cf. Siegel 2008; Clements 2014). The researchers then asked these individuals about their pronunciation when they were reading the Koran. They responded that while reading the Koran, geminates ought to be produced so that the Koran would be read correctly, that the spoken EPA and Koran were ‘two different things’, and that Koranic Arabic ‘is God’s words’, and must not be compromised. For additional validity, eight of the participants who took part in this study were asked to read verses from the Koran that contained geminated consonants. It was revealed that geminated consonants in the Koran were not degeminated by the participants, including speakers of non-geminating L1s.

Second, the participants expressed that they felt degemination represented them as the true speakers of EPA. They could distinguish themselves from local Emiratis who were ‘putting on’ this speech feature while conversing with people from the participants’ community. In addition, the speakers stated that degeminating the consonants gave them a measure of control over the speech rules of the EPA community and reinforced their singularity and solidarity. The participants in the interviews added that this way of speaking Arabic (i.e., without geminates) guaranteed ‘mutual understanding’ between them. It made them be perceived as part and parcel of an already-existing community who spoke EPA with degemination.

Third, those participants who had been in the UAE for 5 to 8 years reported that they would degeminate as they preferred to speak the variety of Arabic (EPA) they learnt from ‘more senior’ workers who had resided in the UAE for more than 20 years. They mentioned that speaking EPA with degemination helped them to ‘market’ themselves, and that degemination identified them as ‘experienced workers in the UAE’. According to their account, Emirati Arabs and Arabs from other nationalities who are in a position to offer employment prefer experienced workers, familiar with the firm and their job in the UAE. Hence, by degeminating some consonants, these speakers mark themselves as being experienced personnel who have resided in the UAE for a long time and are, thus, more familiar with the natives and the work environment. This finding can be accounted for in light of ‘approximation of approximation’ in which those who arrive in a language-contact situation, start speaking the contact language on having contact with members of the same community, i.e., blue collar workers in the UAE who have been working in the country for more than twenty years (See Chaudenson 1992, 2001; DeGraff 2001). Thus, through recursive approximation with these veteran workers, the new ones begin to speak the language variety of former workers.

5. Conclusions

The current study investigated Emirati Pidgin Arabic (EPA) as spoken by unskilled migrant workers of various nationalities in the UAE. Specifically, the current study looked at the reduction of geminates in Emirati Arabic (EA) words in EPA. The results revealed that these workers opt to degeminate the consonant geminates for the following reasons: solidarity and preserving ‘in-group’ dynamics and self-marketing. It was argued that the influence of the mother tongue may not have contributed to degemination; native speakers of languages who do or do not have phonemic geminated consonants, degeminated the target words alike. Of particular interest is the fact that the Muslim subjects in the study (20 of the 25 study participants) reported that they did not degeminate geminate consonants when praying or reciting the Koran in Arabic. It is therefore clear that, first, they had awareness with regard to this aspect of pronunciation, and second, that at least to some degree, they were able to control it in the controlled task given in this study. In other words, it appears that the geminate vs. degeminated consonant realizations form an ‘opt in–opt out’ choice for speakers of EPA.

While we have focused on geminates in individual words in order to streamline our research, we think that the survey should be extended to geminates in longer pieces of natural discourse. This would provide a larger sample and a number of different contexts that would increase the reliability of our conclusions and would uncover whether degemination assumes other functions than that of being a sociolinguistic marker of distinctiveness.