Treball de fi de màster

Màster:

Edició:

Directors:

Any de defensa:

Col⋅lecció: Treballs de fi de màster

Programa oficial de postgrau "Comunicació lingüística i mediació multilingüe"

Departament de Traducció i Ciències del Llenguatge 1

TABLE OF CONTENTS

1. Introduction…………………………………………………………………..…4

2. Theoretical background………………………………………………………....6

2.1.Languages in Lebanon…………………………………………………...….6

2.2.Language choice online………………………………………………..……8

2.2.1. English…………………………………………………………..…..8

2.2.2. MSA…………………………………………………………..…….9

2.3.……………..………………………………………………..10

2.3.1. Romanization as an orthography…………………………………..10

2.3.2. Reasons for Romanization…………………………………………11

3. Informants……………………………………………………...………………14

4. Method…………………………………………………………………...…….15

5. Limitations……………………………………………………...... 17

6. Findings…………………………………………………………………..…....18

6.1.Language choice……………………………………………………..…….18

6.1.1. Emails and formal inquiries……………………………………..….18

6.1.2. Initiating contributions…………………………………………..….18

6.1.3. Responding contributions……………………………………….….19

6.1.4. Whatsapp……………………………………………………….…...20

6.2.Reasons for Romanization…………………………………………….…..22

6.3.The practice of Romanization……………………………………………..22

6.3.1. Phonology

6.3.1.1.Consonants………………………………………………….,...22

6.3.1.2. …………………………………………………….….23

6.3.2. Morphophonology………………………………………………....24 2

6.3.3. Morphology……………………………………………………….25

7. Conclusion…………………………………………………………………….26

8. References…………………………………………………………………….28

3

ABSTRACT

This paper investigates language use online by speakers of Lebanese , focusing in particular on their language choice in online contexts, and the Romanization of this variety of

Arabic. Three speakers with different linguistic backgrounds were interviewed, and samples of their online activities were collected and analysed. The findings show that the language choice of Lebanese Arabic speakers online generally resembles their offline linguistic practices and preferences, with Lebanese Arabic maintaining its informal character online in contrast with English and Standard Arabic, and displaying code-mixing among youth. The results also show that Romanization is inconsistent among speakers, and that one possible factor causing this inconsistency is the speaker's L2.

Keywords: Lebanese Arabic, Computer-mediated communication, Romanization, Language choice

4

1. INTRODUCTION

Computer-mediated communication (CMC henceforth) has been defined as "any communicative transaction that takes place by way of a computer" (McQuails, 2010). A significant amount of the research carried out on CMC focused on language and language use. While the majority of such research revolves around the , several studies have also been carried out on the use of Arabic online (Warschauer et al., 2002;

Palfreyman & Khalil, 2003; Al-Tamimi & Gorgis, 2007; Yaghan, 2008; Daoudi, 2011,

Farrag, 2012).

This paper aims at describing language use on CMC by a small sample of speakers of a specific variety of Arabic, Lebanese Arabic (LA henceforth). In particular, it discusses language choice on CMC by LA speakers in the context of the multilingual Lebanese society, and the practice of LA Romanization ( into the Latin alphabet) as an ad-hoc writing system that was originally developed for lack of supporting software for the , but still prevails although technological advances have now eliminated the technical constraints that prevented its use formerly.

With focus on two forms of CMC (social media website Facebook and mobile messaging application Whatsapp), the paper attempts to examine the contexts of use of one language or another, the reasons for the emergence and persistence of Romanization, and the linguistic aspects of the Romanization practice.

Following an ethnographic approach based on Computer-Mediated Discourse Analysis, defined as "the analysis of logs of verbal interaction" (Herring, 2004), this paper examines language use online by three LA speakers of different linguistic backgrounds, based on interviews and samples of online activities. 5

The paper is structured as follows: a theoretical background on language choice and

Romanization in Lebanon and elsewhere will first be provided, followed by a description of the informants and methodology. The findings will then be presented and analysed.

6

2. THEORETICAL BACKGROUND

2.1.Languages in Lebanon

Lebanese Arabic is a variety of Arabic that belongs to the Levantine dialect, which is one of the four main Arabic dialects.1 It is the first language of some 4.5 million speakers.

As an Arab country, a former French colony, and home to academic institutions and multinational corporations having English as the primary means of communication,

Lebanon’s language scene is distinctly multilingual. This linguistic diversity originated in the period of Ottoman rule (between the fifteenth and twentieth centuries) when European missionaries started establishing schools in the country, then grew bigger during the two decades of French mandate, and later with the global rise of English use as the international lingua franca.2

Moreover, the educational system in Lebanese schools has played a major role in favouring multilingualism. By virtue of educational reforms introduced in 1994, Lebanese schools became constitutionally required to adopt a multilingual approach to learning, with the teaching of the first foreign language (usually French or English) starting at the beginning of schooling, and that of the second starting in the seventh grade at the most,

(StateUniversity.com) with many schools opting to include the teaching of the latter in the curriculum at an earlier level.

This multilingualism is manifested in LA mostly through code-mixing and code-switching, especially popular amongst youth. Moreover, despite the positioning of French throughout the last century as the second language of the majority of LA speakers, English is believed to be on the way to becoming the more dominant foreign language in the country (Shaaban &

Gaith, 1999, Esseili, 2011).

1 The other three are: Mesopotamian Arabic, , and . 2 See Shaaban (1997) for a more detailed historical overview of languages in Lebanon. 7

In addition to having English and French as "culture languages," Lebanon displays a case of diglossia for Arabic, like its Arab neighbours (Warschauer et al., 2002; Ibrahim, Taha, Abu-

Dabbous, & Khatib, 2013). A diglossic situation is one in which a "high variety" of a language coexists with a "low variety", the former usually being reserved to formal and written settings, while the latter is widely used in informal situations and speech (Ferguson

1959, 1972). Thus, the official language of Arab states is the formal, "high variety" of Arabic, i.e. (MSA henceforth), which is used in educational, media, religious, and formal contexts. However, daily communication is carried out almost always in local dialects, which display significant phonological, syntactic, and semantic differences from MSA. Therefore, not only is MSA not spoken as a first language by any Arab population, but it may also be considered a foreign language "to some extent" for speakers of vernacular Arabic varieties (Ibrahim et al., 2013).

The status of MSA as a foreign language for LA speakers is therefore rather plausible, especially taking into account the broad presence of two other foreign languages in the country that are typically learnt in formal educational settings simultaneously with MSA, but with much more focus placed on the former rather than the latter. In fact, in Lebanese schools, the overwhelming majority (77.2%) of the teaching process is done in French or

English, while the minor part (22.8%) is carried out in MSA (Tarazi-Sahab and Moro, 2013).

Nonetheless, it is noteworthy that a large number of the LA speakers who received formal education master MSA fairly well, naturally due to its similarities with LA, but also possibly due to the large exposure they have to MSA in non-educational settings, which may compensate to some degree for its secondary position in education. However, this is obviously not the case for speakers who have not received or completed basic formal education, as their exposure to MSA only through non-educational contexts is not sufficient to establish their good knowledge of the language. 8

2.2.Language Choice Online

Multilingualism provides speakers with several language choices to employ for their communicative purposes at any moment, and this multitude of options extends to online practices on various CMC forms. However, language choice online is not only limited to determining the language(s) internet users opt for among the options available to them according to their linguistic background, but also encompasses the way they "negotiate their choice" in their online exchanges and practices (Lee, 2015, p.120), as well as the technologies and applications available to them in any given context.

2.2.1. English

Previous research on language practices by multilingual individuals online has shown that

English is widely used as the lingua franca among internet users who do not share the same first language (Durham, 2003; Lee, 2015), which shows that language choice online is "a key resource by which to bring together or separate various parts of the networked audience"

(Androutsopoulos, 2014, p.71).

However, in many cases, communities that do share the first language have also been found to conduct their online communication in English. Androutsopoulos (2014) challenges the assumption that communication in English is reserved for international exchanges, arguing that English can also be used within "local" environments to manage the audience and the unfolding of the exchange. A multitude of factors may also lead speakers of the same language to choose to use English, such as their preference for using English for specific topics, contexts, genres, or interlocutors, based on cultural, social, or personal motivations.

For instance, Bianchi (2012) found that English was used by speakers of for sensitive "Western" topics, e.g. homosexuality, considered taboo in their society.

This local communication in English on CMC takes place either by holding the entire communicative exchange in English, or by integrating English into exchanges held in the 9

users’ native language, similarly to code-mixing in spoken language. Previous studies have shown that on CMC, English-only communication is often reserved to formal contexts, while code-switching between English and the native language is observed in informal ones

(Warschauer et al., 2002; Al-Tamimi, A. & Gorgis, D. T., 2007; Seargeant et al., 2012;

Bianchi, 2012; Darwish, 2013). Lee (2007) attributes this tendency to the concept of synchronicity. Informal contexts, e.g. chatting or messaging platforms, are said to be synchronous, as they resemble real-time face-to-face exchanges (where code-switching happens rather naturally) more closely than formal contexts like emails and Facebook status updates, where users can devote more time to the construction of their exchanges and thus do not feel the need to switch codes as much.

This resonates with the findings of Warschauer et al. (2002), which found that formal e-mail communication was carried out in English by the overwhelming majority of young professionals in Cairo, while most informal emails and chats displayed code-switching between English and Arabic. Likewise, in Palfreyman & Khalil (2003), university students in the UAE were found to use English for university-related topics and Arabic for "formulaic" expressions.

Nonetheless, it is important to note that this transition from the native language to English in online contexts, whether partial or complete, does not always reflect language use in offline encounters (Androutsopoulos, 2014). For instance, Seargeant et al. (2012) reported a significant amount of code-switching into English on CMC by Thai speakers, despite the preferred face-to-face language being Thai.

2.2.2. MSA

In the Arab world, CMC users must also make the choice between using MSA or their local dialect. Outside CMC contexts, the vernaculars of Arabic may be found in written form in some contexts that aim at giving forms of written communication a "local flavour", such as 10

local literature, or advertisements on billboards and in written media (Palfreyman & Khalil,

2003). However, this written form of the vernacular does not have any official character or standards.

As for CMC contexts, the usage of the Arabic language is similar to its usage in offline contexts. In other words, the choice is made according to the formality of the exchange: in general, MSA is rarely used in informal contexts compared to vernacular Arabic, in both its

Arabic-script and Latin-script forms (Palfreyman, 2001; Warschauer et al. 2002).3 MSA is usually reserved for formal or special communication, e.g. religious expressions, while vernacular Arabic is used for more trivial, everyday topics (Warschauer et al., 2002; Bianchi,

2012).

2.3.Romanization

2.3.1. Romanization as an orthography

According to Sebba (2009), orthographies are created by assigning characters from a certain script to the words, sounds, or syllables of a particular language according to "conventional correspondences." Importantly, Sebba notes that orthographies do not always have to be standardised, and may often show inconsistencies and variation. This might stem from the unofficial status of the orthography because of its absence from educational curricula

(Androutsopoulos, 2009), or from the lack of a standard writing system for the language itself, such as Cantonese and the spoken Arabic vernaculars.

Romanization as an orthography could therefore be defined as the mapping of the graphemes, , or syllables of a particular language into the graphemes of the Latin alphabet.

The practice of Romanization exhibits much variation between one source language and another, and within the same language as well. For instance, while Romanized Greek uses

3 It should be noted, however, that these studies were carried out at a time when the Arabic script was not as widely available and supported as it is today. For instance, acquiring the on smartphones today is as easy as acquiring the English one. Moreover, most informants in these studies were youth, most of which more proficient in English than speakers of an older age.. 11

either a phonetically-based or an orthography-based approach (see Androutsopoulos, 2009), the relies primarily on -to-grapheme mapping - thus making it closer to being a transcription rather than a transliteration (Palfreyman & Khalil, 2003; Al-

Badrashiny M., Eskander R., Habash N. & Rambow O., 2014). Nonetheless, it also draws on orthographic similarities to characters in the native script as a secondary solution when the former proves to be impossible or inappropriate. Therefore, Arabic phonemes with no Latin graphemic equivalent are sometimes rendered in numerals that look similar to the Arabic

.etc ,<ح> or <7> for ,<ع> or <3> for ,<ء> script grapheme, e.g. <2> for

2.3.2. Reasons for Romanization

It is important to note that Romanization is by no means the result of the technological advances that the last few decades brought about. Besides its official character in some languages, e.g. the Turkish language which started being written in instead of the

Ottoman one by virtue of a 1928 reform, unofficial Romanization schemes have also been noted in the pre-internet era. The practice of rendering Greek in Latin letters was widespread since as early as the nineteenth century for trade, literary, and religious purposes

(Androutsopoulos 2009). Similarly, an (unsuccessful) attempt to render LA in Latin letters was made by nationalist poet Sa’id Akl in the second half of the last century.

However, the advent of the internet contributed largely to the increase of this practice among speakers of languages not written with the Latin alphabet.

The spread of the Romanization phenomenon on CMC can be traced back to the early days of the internet, when computer encoding systems were largely based on the Latin script, forcing those speakers whose native script used non-Latin characters to devise "makeshift" scripts

(Crystal, 2001). Like Greek, Chinese, Japanese, and Russian among others, the Arabic script was originally impossible to use on CMC, because the characters supported by operating 12

systems at the time were limited to the ASCII character set, which consisted of the Latin alphabet, numerals, and some of the most common symbols.

Nonetheless, although technological restrictions might explain the emergence of Romanized orthographies on CMC, they cannot explain their widespread persistence at a time when technological advances, e.g. the development of the character-encoding system, allow for the representation of a much larger number of writing systems. Resorting to the

Latin script despite the availability of the Arabic script was shown in Palfreyman & Khalil

(2003), and Androutsopoulos (2009) explained that, compared to the native Greek script,

Romanized Greek was "so firmly established among early adopters of computer-mediated communication that it was referred to as the ‘old writing method’" (p. 225).

Possible reasons for this were reported to revolve around social, sociolinguistic, and technical factors mostly. Abdel-Ghaffar, N. et al (2011) reported that many Romanize their spoken language to "go with the flow", while Essawi (2011) said that the practice of

Romanization could be explained as an attempt to conform with the social image associated with this use, especially among youth, who believe it makes them look "cool".

From a sociolinguistic point of view, Sebba (2012) establishes a link between language ideologies and identity, stressing the importance of scripts in particular as markers of identity and belonging. In this context, Romanization may connote an attempt to (re)construct a community’s identity so as to establish an affiliation with communities or nations that use the

Latin alphabet. It might therefore be employed to symbolize competence in Western languages, an indicator of high education and prestige (Warschauer et al., 2002; Palfreyman

& Khalil, 2003), or to symbolize an alignment with the modernity of the Latin-alphabet-using

Western world, as in the case of Romanized Greek (Androutsopoulos, 2009).

Multilingual CMC users may also resort to Romanization for practical technological reasons.

Generally, the knowledge of several alphabets gives users the possibility to switch between 13

one alphabet or the other according to the context, practice, or interlocutor; and while this switch is easily done on paper, it is more complicated on CMC (Cassany, 2014). The use of several alphabets at a time online can be a difficult task due to the technological aptitude needed for being familiar with several scripts. Moreover, it can be time-consuming, given the fast-paced nature of online communication, which discourages CMC users from switching from one keyboard to another, especially if they are already familiar with a particular one.

This leads them to resort to the Latin script as a cost-effective solution for time and technological constraints.

14

3. INFORMANTS

Data from the online activities of three LA speakers with different linguistic backgrounds was collected and examined. The informants are:

● G., 53-year-old housewife. She lives in the Greater Beirut area, where she moved

during her teenage years from her hometown in the countryside of Greater Beirut. Her

L2 is MSA, and she has good knowledge of French and basic knowledge of English.

She has been using the internet for a little over a year on her smartphone and never

used computers for online communication. G.'s interactions on Facebook were with

relatives and acquaintances from her hometown and were mostly social formalities,

e.g. congratulating or thanking her Friends, with occasional political posts. Her

interlocutors on Whatsapp were her children who are in their twenties. G. admitted to

preferring sending voice messages on Whatsapp as it is easier than typing messages.

● C., 19-year-old female Translation undergraduate, who lives in the Greater Beirut area

and goes to university there. Her L2 is English, and she has very good knowledge of

MSA and basic knowledge of French. She has been using the internet for almost 12

years. C.'s postings on Facebook were mostly about her social circle and activities,

and her Whatsapp exchanges were held with friends and family, most of whom come

from the same social background as her.

● J. is a 28-year old female public sector employee, with degrees in Journalism and

Political Science. She lives in the Greater Beirut area, and has French as an L2, with

great knowledge of both MSA and English. She has been using the internet for at least

12 years. Her postings on Facebook revolved around her social circle and activities, as

well as social or political issues. Her interlocutors on Whatsapp were relatives who

have the same background as her, as well as colleagues and friends, some of which

come from different linguistic and social backgrounds. 15

4. METHOD

Three sources were used to collect the data: first, interviews were conducted via Skype. A set of questions was prepared beforehand, but during the interviews, the conversation diverged as needed to elaborate or add observations. Notes of the informants' responses were taken.

Informants agreed to be contacted again in case the need for clarification of any point arises.

The interviews had three purposes: (1) to gather general information about the informants’ academic and professional backgrounds, (2) to determine their L2 as well as their linguistic level and capacities in each language4, and (3) to inquire about their linguistic practices both online and offline. One main interview was held with each of the informants, but J. and G. were both contacted after the interview for further clarifications. Second, examples of the informants’ activity on social media website Facebook between January 2015 and May 2016 were collected to examine language choice. The data was divided into initiating and responding contributions, in the terms of Androutsopoulos (2014). Here, the former refer to contributions like status updates, whose audience is the entirety of the user’s "friends" list (or sometimes an even larger network depending on each user’s privacy settings), whereas the latter refer to contributions directed at a much smaller audience, although most times still visible to the larger audience (such as comments and replies). Third, the informants were asked to provide samples of messages sent and received on mobile application Whatsapp that include Romanized LA, regardless of the topic and the interlocutors' quantity and identities.

The interlocutors were asked for consent when possible.

The collected data was then analysed by studying patterns and examples of language use and

Romanization. The corpus amounted to 735 words in 132 turns of conversation from 19

4 Because of several limitations, no attempt to determine the speakers’ L2 and linguistic knowledge by means of a language assessment was made. Rather, informants’ L2 was determined as the language: (1) that they were exposed to the most (besides LA) in the first years of childhood, (2) in which they received most of their education, and (3) in which they consider themselves to be the most dominant overall (on the levels of grammar, speech, and writing). Informants were also asked to rate their productive and perceptive spoken and written skills in each of the three languages from 1 to 10. 16

samples of Whatsapp conversations, and to 18 initiating contributions and 22 responding contributions on Facebook.

17

5. LIMITATIONS

Given the small size of the corpus and the limited number of informants, this study cannot be considered a generalization on language use by all LA speakers, especially that both topics examined in the study, i.e. language choice and Romanization, display significant variability across speakers. Moreover, several individual factors may have affected the results obtained by the three informants. For instance, informant J.'s academic and professional backgrounds can explain her relatively comfortable attitude towards MSA, but might give an inaccurate idea of that of her peers with different backgrounds. On the other hand, informant C. described the MSA education she received at school as "not that good", which adds to the impossibility of generalizing the results observed in this study on Lebanese speakers as a whole.

Nonetheless, the study could serve as a foundation for further research based on patterns of language use found in this study, especially given the lack of literature on LA and its use on

CMC.

18

6. FINDINGS

6.1.Language choice

The informants’ language choice online was shown to mirror the general tendencies observed in the Arab world concerning the formal character of MSA as opposed to the informal character of the vernaculars.

The Lebanese vernacular was mostly represented online using the Latin alphabet, in a code that will be referred to henceforth as RLA (short for Romanized Lebanese Arabic). Code- mixing was frequent in the data of the younger informants, mostly between RLA and English

- something which parallels their speech which often includes code-mixing between LA and

English. Code-mixing was more common among youth than among older LA speakers as evidenced by the informants' responses in the interviews. 5

6.1.1. Emails and formal contexts

The data collected did not cover such formal contexts. However, when asked during the interviews about their language preference for emails, inquiries, and formal contexts, the informants unanimously dismissed the option of RLA, with English-fluent C. and J. opting for English and MSA-dominant G. choosing MSA.

6.1.2. Initiating contributions

In their initiating contributions on Facebook, which are considered less formal than professional emails and similar formal contexts, but more formal than the other contexts examined in this study (responding contributions on Facebook and messaging exchanges on

Whatsapp), C. and J. again behaved differently from G., who only used MSA for her initiating contributions. On the other hand, C. and J. predominantly used English, albeit with the presence of some code-mixing of English and RLA, which will be referred to as CM

5 Here, code-mixing refers to the use of foreign words or expressions when using the Arabic alternative is equally possible. Foreign words or expressions frequently used in LA because the use of the Arabic alternative is unconventional (such as “hotel”, “séchoire”, “online”, “chargeur”) are considered loanwords. Instances displaying them were therefore not considered code-mixing. 19

henceforth (the term "code-mixing" will be used when referring to a different combination of languages). Some initiating contributions were also carried out in MSA, but those were less frequent than in English.

Example (1): Initiating contribution by J. in CM (English and RLA) Mind your own business ya "abouna"! Stop interfering in everything and brainwashing people! [Mind your own business "father" (in reference to priest)! Stop interfering in everything and brainwashing people!]

Example (2): Initiating contribution by C. in English Happy birthday to my much older sister who is now often mistaken for being my much younger sister! I love you to the moon and back ❤

6.1.3. Responding contributions

In general, MSA was rarely used in responsding contributions, while English was relatively more frequent. A significant amount of CM was observed, but with varying degrees between the informants. Unsurprisingly, the amount of code-mixing was higher for J. and C., in accordance with findings of earlier studies on the popularity of this trend among youth.

However, it is unclear whether their better knowledge of foreign languages compared to G. also plays a role in the fewer code-mixing instances in the latter's data.

Example (3): Responding contributions by J. in RLA, French, and English

Friend 1 ➢ J: Hi

J: Ka NAME ka NAME (tagged: Friend 2) khallsine mennik ba2a w ta3e... Aw khallike w ana beje, pour toujours F2: Inte skete wala rah trouhe wala shi!! Yall F1(tagged) soon   J: U never know, yemken tousalo la honik w esta2belkon 3al matar  F2:  nchalla

Translation: F1 ➢ J: Hi J: Oh NAME, oh NAME, (F2) hurry up and come…Or you stay and I come, forever F2: You shut up you won't go!! Come on NAME(F1), soon J: You never know, maybe you will get there and I will be waiting for you in the airport F2: Hopefully

20

Example (4): Responding contribution by G. in RLA

F: y khalikon la ba3d. Happy birthday G: Merci NAME nefra7 menak wy5alilak 5ayak ya Rab

Translation: F: May God preserve you for each other. Happy birthday G: Thank you NAME, may we share your happiness and may God preserve your brother

6.1.4. Whatsapp

Informants’ exchanges on Whatsapp showed more or less similar results to responding contributions in terms of language choice, possibly due to the more or less "private" nature of these interactions.

RLA was extensively used by all three informants (26% of C.'s contributions; 34% for J.; and

86% for G.). Code-mixing (mostly CM, but also between French and RLA) correlated with the age of the informants, as it was heavily present in 19-year-old C.’s exchanges (48%), less so but still significantly in 28-year-old J.’s exchanges (39%), and even less in 53-year-old

G.’s, whose interactions were mostly held in RLA.

The English terms in the CM used in J. and C.’s exchanges were usually related to academic

(elasticity), professional (project, email), or technical issues (monitoring). However, the

English terms in C.’s contributions also revolved around other more trivial issues such as food (salad, appetizer) and days of the week (Thursday, Friday). Meanwhile the little CM found in G.’s exchanges was restricted to formulaic expressions (hi, sorry, please).

Moreover, exchanges by all informants displayed semiotic and para-lexical material, such as laughter expressions and emoticons, which were disregarded from the analysis as they did not pertain to a specific code.

21

Illustrative examples of variability within and between Whatsapp messaging exchanges

Translation:

I: She told me C: When I did an internship there [RLA] C: I benefited a lot [RLA] C: Even if for a month [RLA] I: Really? C: You become more responsible [CM] C: And you become perfectionist and determined [CM] I: Good experience C: Working for them is really nice because they are very straightforward [CM] C: Amazing experience [English] I: I want good experience

Example (5): Whatsapp exchange provided by C., where she uses RLA and CM equally (3 times each) and English once

Translation

G: Where are you just say hello [RLA] I: Hi I: You're still awake G: Yes [RLA] I: How are you? I; We are having a drink G: Fine, cheers, where are you now? [RLA] I: ADDRESS I: (Emoticon) I: If I told you would you know? G: Sorry just checking if you’re at home [CM] I: (Laughter) no need for saying sorry I: No at a pub I: (Sends audio message) G: Ok I know [RLA] G: Excuse me [RLA]

Example (6): Whatsapp exchange provided by G., where she uses RLA only.

22

6.2.Reasons for Romanization

Concerning the reasons for Romanizing LA online, the informants agreed during the interviews that the main reason is that they "got used to it". When asked about their initial reasons for using RLA, informants J. and C., who have a longer history of using CMC than

G., agreed that the use of RLA was the norm on CMC by the time they started using the internet. They noted that because LA is the means used for communication and not MSA, the use of which for regular communicative purposes is "unacceptable" in society, according to

C., the use of MSA (and the Arabic script associated with it) on CMC when Romanization started spreading was considered "weird", according to J. They also cited unfamiliarity with the Arabic keyboard as a possible factor of the initial unpopularity of the Arabic script.

6.3.The practice of Romanization

6.3.1. Phonology

6.3.1.1.Consonants

The Latin characters corresponding to LA consonants are rendered with relative regularity, based on their English or French counterparts. In line with Bjornsson (2010)’s findings, when

English and French differ in the representation of the same sound, informants used the grapheme corresponding to their L2 (or L3 in the case of MSA-dominant G.). For instance, to represent /ʃ/ in the word /ʃu:/ (what), J. and G. followed French orthography, writing [chou], while C. followed English and wrote [shou].

As for consonants that do not exist in English or French, their representations displayed some inconsistencies, as they are either rendered as numerals that are visually similar to the Arabic character, or as the closest-sounding Latin character. For instance, the pharyngeal fricative /ħ/

appears either as the numeral [7] or as its glottal alternative [h]. The general tendency is (ح) towards the use of numerals; nonetheless, C. attributes her limited use of numerals to reasons 23

of aesthetics and "easiness to read by foreigners". When asked about the rules they followed for RLA, informants cited the use of numerals as "conventions".

Table (2): Representation of LA consonants in RLA

/LA/ /LA/ /LA/ /LA/ /LA/

/m/ /tˤ/ /z/ /l/ /w/ <Ø>

/b/ /d/ /zˤ/ /j/ /ħ/ <7>

/f/ /dˤ/ /r/ /k/ /ʕ/ <2> <Ø>

/n/ /s/ /ʃ/ /x/ <5> /ʔ/ <3>

/t/ /sˤ/ /ʒ/ /ɣ/ /h/

6.3.1.2. Vowels

As for vowels, the variation among informants was more remarkable than for consonants, as most sounds had several alternative representations. However, despite the multitude of alternatives, every vowel sound generally had one representation that was the most common, e.g. for /e/ and for /u:/.

Moreover, informants J. and C. were more consistent in their representation of vowels than

G., whose exchanges exhibited the most inconsistency. For instance, the word /mlu:xije/

(name of a meal) was represented by G. in the same sentence in two different ways: mlou5eye and mlou5eyi, where the final vowel /e/ was rendered once as and once as .

Overall, this vowel was represented in the data using 8 different graphemes, 4 of which were exclusively used by G.

G.’s data also shows that she frequently uses the grapheme to represent LA vowel /i/, and the grapheme to represent LA vowel /e/, as in example (5) below. This could be the result of morphophonological influences of MSA, a point discussed more extensively next.

24

Example (7):

G: badek t7ki 3amtek

/baddik teħke ʔɑmtik/

You want to speak to your aunt?

This inconsistency in G.’s representation of LA vowels as opposed to J. and C.’s more consistent representation might be explained in light of the comparison between the vowel systems of the informants’ L2. French and English have extensive vowel systems compared to MSA, which has only three vowels, each having a long and a short form. Therefore,

French and English provide speakers with more phonemes that correspond to LA vowels that do not have an equivalent in MSA, e.g. /ɔ/ or /e/.

6.3.2. Morphophonology

Although G.’s most common representation of /e/ is (like J. and C.), G. also frequently used the graphemes and to represent this sound, in particular when it occurs in the final position (and only then), something which J. and C. never did.

The two graphemes and could be analysed as graphemic realizations of MSA’s long vowel /i:/, frequently found in verb-final and pronoun-final position, as the MSA morpheme

pronounced /i:/, corresponds to a variety of grammatical functions like case, person, and ,[ي] mode.

Another grapheme used by G. to represent was French <é>, mostly in noun-final and adjective-final position, usually occupied in MSA by the grapheme (with the /t/ sound often silent, so the actual realization of the grapheme is /a/). This resorting to French orthography could be explained by the fact that the phonetic oddity that would result from

G.’s rendering of this grapheme according to MSA orthography led her to use the grapheme that her next dominant language provides. Therefore, instead of rendering the LA adjective

/ħɪlwe/ (beautiful) according to MSA orthography <7elwat>, pronounced /ħɪlwat/ (which would sound awkward to LA speakers), she opts for <7elwé>. 25

6.3.3. Morphology

RLA also displayed morphological variation among the informants whose L2 uses the Latin alphabet on the one hand and the one whose L2 is MSA on the other, most notably in the prefixation of some parts of speech. This phenomenon is widely observed in MSA, where several parts of speech are attached to the word they precede. Our focus will be in particular on the definite , usually realized in LA as /l/, the conjunction (and) realized in LA as /w/, and some prepositions which, although not necessarily prefixed in MSA, are sometimes realized in LA as part of the following word due to their reduced form, e.g. MSA

<ʔala> (on) and <ʕila> (to) become <ʔa> and respectively in LA.

The data shows a clear MSA-influenced representation of these elements by G., who frequently attached them to the following word, even when the result sounded awkward or confusing. For instance, G. affixes the definite article to the word (news), resulting in the word la5bar pronounced /laxbar/, which sounds awkward to LA speakers without the clear separation of the two words. In contrast, J. and C. consistently represented these elements as independent words, as in both their L2s. The following example of an exchange by G. illustrates the above-mentioned point:

Example (8):

G: w7lame becheghl lmou5ayam

and-dream of-(the)work of-the-camp6

And dream of the camp’s work.

In contrast, when asked to render the above example in RLA, J. followed the grammar of her

L2, in which the three elements usually affixed in MSA are independent words:

Example (9):

J: w hlame bi cheghel l moukhayam

6 In LA, the possessed is definite but the possessor is not, hence the absence of the definite article after the preposition /be/. 26

7. CONCLUSION

The present paper examined language choice by LA speakers on CMC, and found that it generally mirrors offline language practices, mainly in two aspects: the informal character of

LA in contrast with the formality associated with MSA or English, the multilingual nature of the Lebanese society, as evidenced by the presence of foreign languages and code-mixing, in line with previous studies (Esseili, 2011) that found the use of foreign languages in Lebanon to be highly favoured. A correlation between code-mixing and age could also be drawn from the results: the language of the older generation of LA speakers is mostly dominated by RLA, whereas youth are increasingly expressing themselves with two (and sometimes three) languages. This could shed light on the changes taking place in the current linguistic situation in Lebanon, and puts into question the value of the Arabic language (both MSA and LA) for

LA speakers.

As for the reasons RLA is still prevalent today despite the technological advances that eliminated in large part the need for it, the findings suggest that this is due to sociolinguistic, linguistic, and technical considerations. Romanization is a cost-effective solution for multilingual CMC users who use more than one alphabet, and is believed to have developed into "a style of communication in itself" (Abdallah, 2008). Therefore, RLA being the preferred code of communication online among LA speakers is mainly due to its easiness and familiarity of use.

As for the practice of Romanization itself, Palfreyman (2001) had found that it is highly inconsistent, but that the factors causing this variability were still unclear. The findings of this paper suggest that despite the inconsistency and the absence of official standards or rules for

RLA, phonological and morphological influences of a speaker’s L2 may explain some tendencies observed in RLA. This is in line with the findings of Abdel-Ghaffar, N., et al 27

(2011) on the effect of L2 on the representation of consonants in Arabic vernaculars, and the findings of Bjornsson (2010) on possible morphological influences.

As already mentioned, this study is representative of a small sample of LA speakers, and thus no generalizations can be made. Therefore, an investigation with a bigger corpus and more informants could be carried out to describe RLA more accurately.

An interesting observation throughout the research was the variation between the practices of

G. compared to J. and C. in both language choice and Romanization. Seeing as this variation could be due to several factors, the most probable of which are the script used in the speaker’s L2 and the years of exposure to CMC (and thus RLA), further research could focus on either of these two factors.

28

8. REFERENCES

 Abdallah, S. (2008), Online Chatting in Beirut: Sites of Occasioned Identity- Construction. Ethnographic Studies, 10, 3-22.  Al-Badrashiny M., Eskander R., Habash N. & Rambow O. (2014). Automatic Transliteration of Romanized Dialectal Arabic. CoNLL.  Al-Tamimi, A. & Gorgis, D. T. (2007) ‘Romanised Jordanian Arabic e-messages’. The International Journal of Language Society and Culture, (21). Available from http://www.educ.utas.edu.au/users/tle/JOURNAL (Accessed 02 June 2014).  Androutsopoulos, J. (2009), ‘Greeklish’: Transliteration practice and discourse in a setting of computer-mediated digraphia. In A. Georgakopoulou and M. Silk (eds.) Standard Language and Language Standards. Farnam: Ashgate. 221–249.  Androutsopoulos, J. (2014), Languaging when contexts collapse: audience design in social networking. In Discourse, Context & Media (2014) Special Issue on Digital language practices in superdiversity. In press, available at ://dx.doi.org/10.1016/j.dcm.2014.08.006  Bianchi, R. M. (2012) Glocal Arabic online: The case of 3arabizi. Studies in Second Language Learning and Teaching, [S.l.], vol. 2, no. 4, p. 483-503. ISSN 2084-1965. Available at: http://pressto.amu.edu.pl/index.php/ssllt/article/view/5132/5211. Accessed: 02 June 2016. doi:http://dx.doi.org/10.14746/ssllt.2012.2.4.3.  Cassany, D. (2014) “Apropiación y uso del alfabeto español”, Journal of Spanish Language Teaching, 1:1, 31-45. ISSN: 2324-7797; DOI: 10.1080/23247797.2014.898499  Crystal, D. (2001). Language and the Internet. Cambridge: Cambridge University Press.  Daoudi, A. (2011). Globalisation and e-Arabic: The emergence of a new language at the literal and figurative levels. Studies In Slavic & General Linguistics, 3861-76.  Darwish, K. (2013). Arabizi detection and conversion to Arabic. CoRR abs/1306.6755  Essawi, R. (2011), “Arabic in Latin Script: Who is using it and why in the Egyptian Society.” In A. AlIssa, & L. S. Dahan, Global English and Arabic Issues of Language, Culture, and Identity. Peter Lang Publishers.  Esseili, F. (2011), English in Lebanon: Implications for national identity and language policy. Ed.D. Dissertation, Purdue University.  Farrag, M. (2012), “Arabizi: a writing variety worth learning? an exploratory study of the views of foreign learners of Arabic on Arabizi.” MSc Thesis, American University in Cairo, Arabic Language Institute, Cairo, .  Ferguson, C. (1959). Diglossia. Word, 15, 325–340.  Ferguson, C. (1972). Diglossia. In Giglioli, P. (Ed.), Language and social context (pp. 232–251). Harmondsworth , UK : Penguin.  Herring, S. C. (2004), Computer-mediated discourse analysis: An approach to researching online behavior. In Barab, S. A., Kling, R. & Gray, J. H. (Eds.), 29

Designing for Virtual Communities in the Service of Learning (pp. 338-376). New York: Cambridge University Press.  Ibrahim, R., Taha, H., Abu-Dabbous, A. & Khatib, A. (2013), Electronic Reading and Writing in Spoken and Written Arabic: A Case Study. Theory and Practice in Language Studies vol. 3, no. 9, pp. 1497-1508.  Lee, C. (2007), Linguistic features of email and ICQ instant messaging in Hong Kong, in B. Danet and S. C. Herring (eds.), The multilingual Internet: language, culture, and communication online, Oxford University Press: New York, pp. 184– 208.  Lee, C. (2015), Multilingual resources and practices in digital communication", in The Routledge Handbook of Language and Digital Communication ed. Alexandra Georgakopoulou and Tereza Spilioti . (Abingdon: Routledge, 27 Jul 2015 ), accessed 06 Jun 2016 , Routledge Handbooks Online.  McQuail, D. (2010), McQuail's mass communication theory. Sage publications.  Palfreyman, D. (2001). Keeping in touch: Online communication and tradition in a Gulf Arab context. Unpublished paper.  Palfreyman, D. & Al-Khalil, M. (2003), “A Funky Language for Teenzz to Use:” Representing in Instant Messaging. Journal of Computer-Mediated Communication, 9: 0. doi:10.1111/j.1083-6101.2003.tb00355.x  Seargeant, P., Tagg C. & Ngampramuan W. (2012). Language choice and addressivity strategies in Thai-English social network interactions. Journal of Sociolinguistics 16: 510–531.  Sebba, M. (2009). Sociolinguistic approaches to writing systems research. Writing Systems Research 1/1: 35–49.  Sebba, M., Jaffe, A., Androutsopoulos, J., & Johnson, S. (Eds.) (2012). Orthography as Social Action: Scripts, Spelling, Identity and Power. (Language and Social Processes). Berlin: De Gruyter.  Shaaban, K. (1997), Bilingual education in Lebanon. In J. Cummins & D. Corson (eds) Encyclopedia of Language and Education: Bilingual Education, vol. 5, pp. 215– 59. Dordrecht: Kluwer Academic.  Shaaban, K. & Gaith, G. (1999), Lebanon’s language-in-education policies: From bilingualism to trilingualism. Language Problems and Language Planning 23, 1 – 16.  StateUniversity.com, Lebanon - Constitutional Legal Foundations - Education, Language, Educational, and Lebanese - Available at: http://education.stateuniversity.com/pages/826/Lebanon-CONSTITUTIONAL- LEGAL-FOUNDATIONS.html  Tarazi-Sahab L. & Moro M. (2013). Changer de langue à l'adolescence chez les adolescents bilingues au Liban. Une modalité de séparation-individuation. In La psychiatrie de l'enfant 2/2013 (Vol. 56), p. 393-410. Available at: www.cairn.info/revue-la-psychiatrie-de-l-enfant-2013-2-page-393.htm.  Warschauer, M., El-Said, G. R. & Zohry, A. G. (2002), Language Choice Online: Globalization and Identity in Egypt. Journal of Computer-Mediated Communication, vol. 7, no. 4, available at: http://onlinelibrary.wiley.com/doi/10.1111/j.1083- 6101.2002.tb00157.x/full 30

 Yaghan, M. A., (2008), Arabizi: A Contemporary Style of Arabic Slang. Design Issues, 24 (2), 39-5.