STYLE SHIFTING IN EGYPTIAN AND TUNISIAN ARABIC:

A SOCIOLINGUISTIC STUDY OF

MEDIA ARABIC

by

Casey Michelle Faust

A thesis submitted to the faculty of

The University of Utah
in partial fulfillment of the requirements for the degree of

Master of Arts

in

Middle East Studies: Arabic and Linguistics

Department of Languages and Literature

The University of Utah

December 2012
Copyright © Casey Michelle Faust 2012

All Rights Reserved

 ABSTRACT

This thesis investigates speech accommodation and dialect leveling in three

episodes of the Al-Jazeera program ‫ ﻤﻣﻔﺘﻮﻭوﺡح ﺤﺣﻮﻭوﺍاﺭر‬ḥuwār mɛftūḥ “Open Dialogue”, with

particular focus on the phonological change of /ḍ/ > [ð̣] (or ‫)ﻈﻇﻆﻅظ > ﺾﺽض‬, so that a word

like ‫ ﺍاﻳﻀﺿ ًﺎ‬/ʔaɪjɪḍan/ ‘also’ > [ʔaɪjɪð̣an] in the Tunisian dialect. This study also looks at

the phonological change of ‫ ﻈﻇﻆﻅظ‬/ð̣/ > [ẓ] in the Egyptian dialect, as well as lexical and

syntactic differences between the use of relative pronouns and particles of negation.

The episodes examined vary in their inclusion of speakers from across the

Arabic-speaking world, and cover a range of speaking styles from reading to debating,

to panel discussions, and street interviews. This thesis posits that Arabic speakers

reduce dialect differences when interacting with others not familiar with their dialect,

illustrating how Arabic speakers strike a balance between the mutually comprehensible

“standard” and their dialect inclinations.

While the Egyptian panel maintains both phonological and lexical characteristics

of their dialect, the in-studio Tunisian guests predominantly use the standard language.

However, there are significantly more dialect features in the speech of on-the-street
Tunisians. Based on the data set, the Egyptians are able to maintain their dialect in the

media setting because it is widely understood throughout the Arab world. Since the

Tunisian dialect is not as commonly understood, the Tunisian studio guests use the

standard to reach a pan-Arab audience.

This sociolinguistic study illustrates the complexities of how Arabic-speakers

manipulate their language depending on the social context and their audience and

challenges the notion of diglossia. Furthermore, this thesis provides a description of

some characteristics of Tunisian Arabic, which has not been well studied in the

literature.

iv
 TABLE OF CONTENTS

ABSTRACT……………………………………………………………………………………….iii

LIST OF FIGURES………………………………………………………………………………vii

LIST OF TABLES……………………………………………………………………………….viii

1: INTRODUCTION...........................................................................................................1

2: LITERATURE REVIEW………………………………………………………………………..4

Diglossia………………………………………………………………………………….5
Levels of Arabic………………………………………………………………………..12
Educated Spoken Arabic………………………………………………………………13
Code Switching………………………………………………………………………...17
Style Shifting: A New Approach……………………………………………………...23
Conclusion...........................................................................................................26

3: THE PROGRAM AND LINGUISTIC VARIABLES………………………………………….29

The Media Context…………………………………………………………………….30

The Program……………………………………………………………………………31
The Episodes……………………………………………………………………………33
Variables.............................................................................................................35
Phonological............................................................................................35
Lexical and Syntactic...............................................................................37
Negation......................................................................................37
Relative Pronouns……………………………………………………38
Other Variables........................................................................................38
Conclusion……………………………………………………………………………...39
4: METHODOLOGY…………………………………………………………………………….40

Data Collection…………………………………………………………………………40
Categorization of Speakers: Speech Setting and Interlocutor..............................43

5: RESULTS AND DISCUSSION.......................................................................................49

Phonological Variable ‫( ﺾﺽض‬ḍ) > ‫ ﻈﻇﻆﻅظ‬/ð̣/................................................................49

The Host..................................................................................................49
The Guests...............................................................................................55
Phonological Variable: ‫( ﻈﻇﻆﻅظ‬ð̣) > [ẓ]....................................................................59
Statistical Analysis..............................................................................................62
Lexical and Syntactic Variables...........................................................................63
Negation..................................................................................................63
Relative Pronouns....................................................................................66
Advantages..........................................................................................................70
Limitations..........................................................................................................71
Conclusion...........................................................................................................73
A Closer Look at Diglossia and Code Switching .................................................74

6: CONCLUSION.............................................................................................................81

Future Research...................................................................................................81

APPENDIX: TRANSCRIPTION SYSTEM...........................................................................83

REFERENCES..................................................................................................................85

vi
 LIST OF FIGURES

FIGURE PAGE

1. Token spreadsheet……………………………………………………………………..41

2. (ḍ) and (ð̣) spectrogram……………………………………………………………….41

3. Host (ḍ) variation……………………………………………………………………...51

4. Guests (ḍ) variation.............................................................................................57

5. Egyptians (ð̣) variation………………………………………………………………..61

6. Particles of negation variation……………………………………………………….66

7. Relative pronoun variation…………………………………………………………...69

 LIST OF TABLES

TABLE PAGE

1. Categorization of Guests………………………………………………………………44

2. Counts of tokens of (ḍ) produced by host…………………………………………..50

3. Counts of tokens of (ḍ) produced by guests………………………………………...57

4. Counts of tokens of (ð̣) produced by host and guests……………………………...60

5. Counts of negation particles produced by host and guests………………………..65

6. Counts of relative pronouns produced by host and guests………………………..67

 CHAPTER 1

INTRODUCTION

Language use is perhaps one of the most interesting aspects of human

interaction. How do individuals change the way they speak? Is it the audience, the

setting, or the conversation topic? Theses are the sort of questions that sociolinguistics

endeavors to answer by taking factors like age, gender, socioeconomic status, and

education as variables to investigate linguistic variation among individuals and groups.

Linguistic variation is usually examined from a standard/nonstandard paradigm

that can entail certain value judgments towards each variety. The standard is usually

formal, educated, proper, whereas the nonstandard is informal, uneducated, and casual.

The variationist framework sought to remove this value judgment and simply report on

how language is actually being used. At the same time, this approach tries to explore

attitudes towards certain variants where possible in order to understand the social

context in which variation occurs. These studies give us insight into sound changes in

progress, new words entering the language, and different grammatical structures in use.
 2

In the variationist framework proposed by Labov (1963), sociolinguists seek to

understand “the social motivation” for change.

As will be discussed in Chapter 2, linguistic variation in Arabic has been

dominated by the theory of diglossia. While diglossia separated the standard and

nonstandard varieties as completely distinct, the notion of an Arabic continuum with

various levels between them sought to conceptualize the language as more dynamic. I

will also discuss the idea of a codified third language between the standard and

nonstandard varieties spoken by educated speakers. Since neither of these theories

solved the theoretical issues of diglossia, the code-switching framework has also been

proposed to explain how speakers can switch between different varieties. Finally, I will

discuss how the style-shifting framework is better suited for an investigation of

linguistic variation in Arabic.

This paper is a sociolinguistic study of media Arabic used on an Al-Jazeera talk

show. The media is an interesting context in which to examine linguistic variation

because it provides a public forum for debating current events and issues that affect

people’s everyday lives. Al-Jazeera in particular will be an interesting context, because

as Lynch (2007) writes, “while Al-Jazeera has faced mounting competition, it remains

the one station watched by virtually everyone, making its programs the ‘common
 3

knowledge’ of Arab politics, which all Arabs can reasonably assume that others have

seen and are prepared to discuss” (p. 103). In this study, I analyze linguistic variation

on three episodes of the program ‫ ﻤﻣﻔﺘﻮﻭوﺡح ﺤﺣﻮﻭوﺍاﺭر‬ḥuwār mɛftūḥ ‘Open Dialogue’. Note that

Arabic words in this paper are transcribed using IPA. For a complete list of my

transliteration system, please refer to the appendix. The episodes contain speakers from

a variety of dialect groups, as well as a variety of speech settings. The main topic of the

episodes is the Arab Spring, with particular focus on the uprisings in Egypt and Tunisia.

First, I examine variation in the phoneme ‫( ﺾﺽض‬ḍ), which in Tunisian Arabic (TA)

is often pronounced as ‫[ ﻈﻇﻆﻅظ‬ð̣], whereas in Egyptian Arabic (EA) it is sometimes

pronounced as [ẓ]. Second, I analyze variation in the phoneme ‫( ﻈﻇﻆﻅظ‬ð̣), which is often

pronounced as [ẓ] in Egyptian Arabic. Next, I examine syntactic and lexical variation by

focusing on particles of negation and the relative pronouns in Egyptian and Tunisian

Arabic. These variables illustrate style shifting on the phonological, lexical, and

syntactic levels of the language, which will highlight how speakers from various dialects

switch between various styles at their disposal in different settings.

 CHAPTER 2

LITERATURE REVIEW

In this chapter, I examine major theoretical issues regarding the sociolinguistic

situation of Arabic. I focus on the concept of diglossia and the theoretical ramifications

of this preliminary theory. Levels of Arabic, the concept of a codified middle language

called Educated Spoken Arabic, as well as the frameworks of code switching and style

shifting will each be discussed in turn.

It should be noted that this literature review is not a comprehensive account of

the complexity of sociolinguistic issues related to Arabic. As mentioned earlier, several

factors influence linguistic variation, and some of those variables like gender and

education are outside the scope of this study. Additionally, this study is not an

exhaustive examination of linguistic variation in these episodes. The variables

investigated here were chosen because of their salience in the data set, and will offer a

survey of how Arabic speakers use differ markers of standard and nonstandard variants.
 5

Diglossia

When students of Arabic start studying the language, they are often

overwhelmed by the differences between the standard fuṣḥā (‫ )ﺍاﻟﻔﺼﺻﺤﺣﻰ‬and colloquial

varieties. They are told that the Arabic linguistic situation is diglossic, which Ferguson

(1959) defined in his seminal article as:

A relatively stable language situation in which, in addition to the primary

dialects of the language (which may include a standard or regional standards),
there is a very divergent, highly codified (often grammatically more complex)
superposed variety, the vehicle of a large and respected body of written
literature, either of an earlier period or in another speech community, which is
learned largely by formal education and is used for most written and formal
spoken purposes but is not used by any sector of the community for ordinary
conversation. (p. 16)

This definition of the Arabic linguistic situation has dominated the field of Arabic

linguistics for the past 50 years. Ferguson was trying to explain the apparent disconnect

between the formal fuṣḥā, the High (H) variety derived from Classical Arabic, and the

informal, colloquial (ʕāmijjə), Low (L) varieties that are markedly different from fuṣḥā

and each other.

Ferguson claimed “one of the most important features of diglossia is the

specialization of function for H and L” (p. 235). Diglossia predicts that fuṣḥā, the (H)

variety, will be used in specific domains like the mosque, university, broadcast media,

newspapers, and literature, whereas ʕāmijjə, the (L) variety, will be used in
 6

conversations with family and friends or members of lower or working classes, radio

soap operas, and captions on political cartoons. According to this theory, Arabic

speakers will use features of one variety or the other, “with the two sets overlapping

only slightly” (p. 235-236).

More broadly, fuṣḥā will be spoken in formal, institutional domains whereas

ʕāmijjə will be spoken in informal, mundane domains. An example will illustrate this

functional differentiation. If an Egyptian Arabic speaker wanted to order a cup of coffee

at a cafe, an informal domain, they will use ʕāmijjə by saying:

(1 ) ‫ﻋﺎﻳﺰ‬ ‫ﻗﻬﮭﮫﻮﻭوﺓة‬ ‫ﻟﻮﻭو ﺴﺳﻤﻣﺤﺣﺖﺕت‬

ʕaɪjɪz ʔahwə law sɛmḥat

'I want coffee please.'

If this same speaker were to order in fuṣḥā, they would say:

(2) ‫ُﺃأﺭرﻳ ُﺪﺩد‬ ‫ﻗﻬﮭﮫﻮﻭوﺓة‬ ‫ﻤﻣﻦﻥن ﻓﻀﺿﻠﻚﻙك‬

ʔu-rīd-u qahwə mɪn fəḍlɪk

'I want coffee please.'

To an outside observer, these two statements seem like they come from completely

different languages. However, the theory of diglossia would predict that the second
 7

example would never be uttered due to the informal domain of the cafe, which would

entail the L variety.

In this example, we can see that Ferguson was correct in a number of respects.

There are certain features of fuṣḥā, whether lexical, morphological, or syntactic, that are

not found in the spoken varieties and vice versa. Lexically, the verb for ‘I want’ is

completely different in both varieties (ʕaɪjɪz vs. ʔu-rīd-u). Therefore, one will only hear

ʔu-rīd-u in a formal domain like a political speech instead of the more informal ʕaɪjɪz.

This example is just one of several lexical items where there is a clear choice between

using a fuṣḥā word versus one from ʕāmijjə. Sometimes linguists will refer to clear

choices like this as a “diglossic” choice, because the lexical item exists in one variety

but not the other.

However, the situation becomes more complex on the phonological level where

Arabic speakers can pronounce a fuṣḥā word, but with their dialect phonology, as in the

second word in this example. For example, Egyptians often pronounce ‫ ﻕق‬/q/ as [ʔ]

(Bahloul, 2007; Haeri, 1996; Soliman, 2008). Therefore, in the ʕāmijjə version of this

statement, the /q/ in the word qahwə was glotallized to [ʔ]. In the literature on

diglossia, the trend has been to categorize ʔahwə as ʕāmijjə because of this sound

change. However, we must question if it is theoretically sound to separate these

 8

phonological variants of the same word (qahwə and ʔahwə) as belonging to completely

distinct varieties.

Let us look at another example from Egyptian Arabic. Since Egyptians pronounce

the phoneme ‫ ﺝج‬/ʤ/ as [g], instead of saying [ʤejš] for the word ‫' ﺠﺟﻴﺶﺵش‬army', Egyptians

would say [gejš]. This phonological difference is so ingrained in the Egyptian dialect

that Egyptians will maintain /ʤ/ > [g] even when they are speaking fuṣḥā in formal

domains. When former President Hosni Mubarak would give formal political speeches,

he would read his speeches mostly in fuṣḥā, but maintained the /ʤ/ > [g] variation. In

Bassiouney’s (2009) sociolinguistic analysis of a political speech by Mubarak, she gives

the following excerpt:

fa qaḍiyyat-ū ṣ-ṣādirā-tī l-miṣriyya qaḍiyyat-ūn maṣīriyyah

Thus, issue-nom det-exports-gen det-Egyptian issue-nom crucial

yaʤ[g]ib ʕan tašgal-ā ʔhtimām-ā kull-ū li-fiʔāt allati

must that 3fsg-occupy-sub interest-acc all-nom det-people rel

tataḥammal-ū ʤ[g]ānib-ān min ʕibʔ wa masʔūliyyat-ī

3fsg-carry-ind part-acc from burden and responsibility-gen

l-intāʤ[g]-ī fi maṣr
det-production-gen in Egypt

‘This is because the issue of Egyptian exports is a crucial issue that has to occupy the

minds of everyone who is involved in Egyptian production’ (Bassiouney, 2009, p. 75).

 9

In this example, I have bolded the case endings that are a distinctive feature of fuṣḥā,

and we can see from the brackets that Mubarak uses the ʕāmijjə [g] variant consistently.

This example should make us question whether the theory of diglossia can be supported

by data from actual speech, because here it is not clear if the functional differentiation

of the two varieties is categorical in nature.

A crucial question here is whether this slight phonological change is so dramatic

that it makes the entire morpheme belong to one variety and not the other. Throughout

the literature on Arabic linguistics, the answer has been yes; this morpheme is now

categorized as ʕāmijjə and not fuṣḥā. However, do other linguists adopt the same

approach with nonstandard variants in other languages?

Let us look at an example from American English, where a phonological change

known as t-glottalization, resulting in /t/ being pronounced as [ʔ], has received recent

attention (Eddington & Channer, 2010 in Western dialects; Eddington & Taylor, 2009;

Roberts, 2006 in Vermont). One phonetic environment in which t-glottalization occurs

in most varieties of American English is intervocalically, so that a word like /mawntɪn/

is pronounced as [mawʔɪn]. This feature is so prominent in my own dialect that I will

probably say [mawʔɪn] even in formal domains like a job interview. While this

variation between [t] and [ʔ] is stigmatized in certain areas, I doubt the interviewer
 10

would gasp in shock that I would use a dialectal variant in such a formal domain.

These examples from Arabic and English illustrate a crucial theoretical question

of how we handle linguistic variation. Mubarak and I are both using informal (L)

phonemes where the formality of the domain implies that the standard is expected.

However, while my speech would probably be interpreted as spoken American English,

Mubarak’s speech is interpreted as switching between two diglossic codes. Do not all

languages have standard and nonstandard variants?

Additionally, Mubarak’s speech poses a methodological question of how to

categorize each morpheme as belonging to one code and not the other. As I mentioned

earlier, one of the striking features of Mubarak’s political speech was his use of fuṣḥā

case endings, which are not present in ʕāmijjə. How then, do linguists use diglossia to

categorize a lexical item like ʤ[g]ānib-ān ‘part’? Here, Mubarak is using the [g] (L)

variant, but with the fuṣḥā (H) accusative case ending –ān. A similar example occurs a

few seconds later when he says l-intāʤ[g]-ī, using the [g] variant with the fuṣḥā genitive

case ending. These two examples are what are referred to in the literature as

‘intermediate forms’ (Bassiouney, 2009; Boussofara-Omar, 1999; Eid 1988). They

possess features from both varieties that are impossible to qualify as belonging to one

variety and not the other, and pose serious theoretical and methodological
 11

consequences that will be discussed in more detail below.

Similar to the issue of intermediate forms is the question of what to do with

items that are shared between both varieties. For example, if an Arabic speaker wants to

say ‘I have’, they can say ‫ ﻋﻨﻧﺪﺩدﻱي‬ʕnd-ī. This phrase is phonologically, morphologically, and

syntactically identical in both varieties. How, then, is it categorized? Some linguists

have classified these items as ‘shared’ (Boussafara-Omar, 1999; Eid, 1988), but they are

often ignored in sociolinguistic studies, despite their pervasiveness in the language.

These examples show the inadequacy of the diglossic framework to explain

language use in Arabic. Such a black and white view of linguistic variation fails to

explain the presence of L (ʕāmijjə) features in H (fuṣḥā) domains, and H features in L

domains, and does not explain how to handle intermediate forms. Ferguson (1991)

admitted that his original article described a vague linguistic situation that was

supposed to be the starting point for a discussion and not a prescriptivist framework.

Despite criticism of the concept and a proliferation of theories seeking to refine it,

diglossia's legacy permeates the literature in Arabic linguistics, where fuṣḥā is placed at

one end of the theoretical spectrum and ʕāmijjə at the other, with little discussion of the

interaction between the two.

 12

Levels of Arabic

In the next few sections, I will discuss various theories that have been proposed

to fix the theoretical and methodological issues posed by the original diglossic

framework. The theory of diglossia invokes an image or metaphor of two separate boxes

for two completely distinct language varieties: fuṣḥā and ʕāmijjə. However, since

language use is not so black and white, Badawi (1973) suggested a different metaphor:

a continuum where each variety is placed at either end, with various levels between

them. Badawi’s theory of levels sought to address the fact that Arabic speakers can use a

combination of features from either variety when they speak. He identified five major

levels:

1. fuṣḥā at-turāθ: ‘heritage classical’

2. fuṣḥā al-ʕaṣr: ‘contemporary classical’
3. ʕāmijjət al-muθəqqəfīn: ‘colloquial of the cultured’
4. ʕāmijjət al-mutanawwɪrīn: ‘colloquial of the basically educated’
5. ʕāmijjət al-ʔumijīn: ‘colloquial of the illiterates’

At the fuṣḥā at-turāθ level, we are to assume that this speech will be 100% fuṣḥā,

whereas at the ʕāmijjət al-ʔumijīn level, we assume that this speech will be 100% ʕāmijjə,

reinforcing the diglossic separation of the two at opposite ends of the spectrum. The

fuṣḥā al-ʕaṣr could perhaps be 70% fuṣḥā and 30% ʕāmijjə, and the rest of the varieties

would be different combinations of percentages of each variety.

 13

However, there are a number of issues with this framework. In particular, it is

not clear what the levels theory would predict of actual language use. What places a

speaker on a particular level? How does one assess to which level a specific utterance

belongs? The levels metaphor is descriptive, but impossible to quantify or predict.

Badawi admits that there can be infinite levels, and each Arabic speaker has varying

access to more or one of these levels. Bassiouney (2009) points out that “it is not clear

whether the colloquial levels are built on socioeconomic variables like education or just

‘stylistic registers’, or whether they can be both” (p. 15). While the levels metaphor

gave a more fluid view of language use by recognizing the reality of speakers using

features from both varieties, it is not clear what, if anything, governs the relative

distribution of features. Therefore, it represents an important step in the literature, but

not a paradigm shift.

Educated Spoken Arabic

So far, we have seen that the separate boxes (diglossia) theory and the

continuum (levels) theory failed to capture the complexities of language use in Arabic.

Another theory that tried to capture this phenomenon is the idea that there is a codified

third language between the standard and spoken varieties, which is spoken by educated

Arabs in situations of interdialectal contact. This third language is called Educated

 14

Spoken Arabic (ESA), and in this framework, Arabic speakers, instead of speaking

purely formal fuṣḥā, will interject various features of their dialect that they believe are

mutually intelligible. The theory of ESA predicts that educated Arabic speakers will

employ fuṣḥā vocabulary, perhaps with dialectal phonology and certain lexical items

that are shared in most of the dialects, like the ʕāmijjə word ‫ ﻛﻮﻭوﻳﺲﺱس‬kwejs ‘good’ instead of

fuṣḥā ‫ ﺠﺟﻴﺪﺩد‬ʤaɪjid.

This phenomenon is connected to the theory of linguistic leveling, which occurs

when dialectal differences are reduced (leveled) for various reasons. Linguistic leveling

is related to issues of prestige, which is particularly prominent in Arabic given that

traditionally, the dialects were stigmatized as “not Arabic” and “mistakes”, while fuṣḥā

was held in high regard as the language of religion, education, and politics. Therefore,

Arabic speakers may avoid features of their local dialect in order to sound more

educated and gain more prestige.

However, it should be noted that prestige is relative and there can be various

attitudes towards a particular dialect within a country or geographical region

(Bassiouney, 2008; Haeri, 1996; Ibrahim, 1986; Miller, 2004). Most notably, the

Egyptians are quite proud of their dialect, which is related to the nationalistic

movement of Gamal Abdul Nasser and the glorious past of ancient Egypt. Therefore,
 15

regional dialects can acquire a local prestige. This is not to say that fuṣḥā is not

prestigious in Egypt, because it is still the language of religion, education, and politics. I

merely intend to point out that local, regional, and community attitudinal

considerations must be taken into account when addressing the issue of prestige.

Linguistic leveling is also related to comprehension issues, particularly when

large geographic distances separate the dialect areas. For instance, Moroccan Arabic is

often given as an example of a dialect that is perhaps farthest from fuṣḥā because of

influences from French and Berber phonetically, morphologically, and syntactically.

Therefore, the theory of ESA would predict that a Moroccan speaker may be very likely

to level features of his dialect in interdialectal conversation and use more fuṣḥā features

in order to ensure mutual comprehension. The issue of comprehensibility is less of an

issue for Egyptians due to the popularity of Egyptian cinema and soap operas, which are

popular throughout the Arab world.

While there are several articles discussing ESA as a theory (El-Hassan, 1977;

Meiseles, 1980; Mitchell, 1978, 1986), there are far fewer sociolinguistic studies of ESA

in use. One exception is Sallam’s (1980) study, which showed that educated Lebanese

speakers from Beirut, who usually pronounce the fuṣḥā phoneme ‫ ﻖﻕق‬/q/ as ‫[ ﻚﻙك‬k], used

markedly more [q] variants in conversation with other Arabic speakers from different
 16

countries, meaning that the Lebanese leveled their local variant in interdialectal

conversation.

I think this framework of ESA is correct in trying to identify common features

that seem to be codified in a ‘third’ language, which is closely related to Ferguson's

(1959) proposal of the Arabic koine. Ferguson described the koine as a form of the

language that is

A relatively homogenous koine not based on the dialect of a single center,

[that] developed as a conversational form of Arabic and was spread over most of
the Islamic world in the first centuries of the Muslim area....This koine existed
side by side with the ‘Arabiyyah although it was rarely used for written
purposes, and...most modern dialects, especially those outside Arabia are
continuations of this koine. (p. 51)

He identified 14 features of the koine, such as the loss of the glottal stop and the dual,

as well as the use of the ʕāmijjə verb ‫ ﺸﺷﻮﻭوﻒﻑف‬/šūf/ instead of the fuṣḥā ‫ ﺭرﺍاﻯى‬/raʔā/. These are

still true of the Arabic dialects today; however, the vast geographical diversity of the

Arabic dialects makes the koine difficult to define.

A few studies of interdialectal conversation have shown that speakers from

various regions will reduce or level their dialectal differences in order to facilitate

communication between speakers from different geographical areas (Bahloul, 2007;

Sallam, 1980), and I do think this is a particularly prevalent phenomenon in the media

context for reasons that will be elaborated below. However, the biggest problem with
 17

the idea of Educated Spoken Arabic (ESA) is that its characteristics are largely

undefined, and few studies have detailed its intricacies.

Code-Switching

The code-switching framework has also been suggested as an explanation for

linguistic variation in Arabic, where speakers are seen as switching between standard

and colloquial codes. Many and perhaps the majority of code-switching studies focus on

switches between distinct languages. In fact, there are several studies that investigate

the rules for switching between Arabic (whether fuṣḥā or ʕāmijjə) and French due to the

prevalence of Arabic/French bilingualism in North Africa. For example, Belazi (1991),

Lawson and Sachdev (2000), Bouzemmi (2005), Baoueb (2009), and Sayahi (2011)

have examined this phenomenon in Tunisia alone. Studies of such obvious code

switching are perhaps more convenient because it is easier to classify a morpheme as

belonging to either Arabic or French, and “Arabic” is usually used as a catchall term

that avoids the problem of having to classify them as fuṣḥā or ʕāmijjə.

However, there are far fewer studies on so-called “diglossic code-switching”

between fuṣḥā and ʕāmijjə varieties of Arabic. This is most likely due to the theoretical

and methodological issue mentioned earlier of dealing with the intermediate forms

because it becomes much harder to classify a morpheme as belonging to only one of the
 18

varieties. The weakness of the code-switching framework is that it emphasizes the

separation of fuṣḥā and ʕāmijjə as distinct varieties, and usually does not address the

issue of the intermediate forms. For example, Eid (1988) admits that the intermediate

forms pose a methodological problem because they cannot provide evidence for or

against switching. Because the goal of her study is to examine where clear switches

between standard and Egyptian features occur, Eid’s solution is to throw these examples

out of the analysis entirely, making the code-switching approach much easier to apply.

However, throwing out the intermediate forms entails ignoring a great bulk of data that

is gathered on spoken Arabic, as will be shown later, and we must address how to

handle them theoretically and methodologically.

Boussofara-Omar's (1999) dissertation applies Myers-Scotten's (1993) Matrix

Language Frame (MLF) model to code switching between fuṣḥā and ʕāmijjə in the

speeches of the former president of Tunisia Habib Bourguiba. Her dissertation is an

exception from the other code-switching studies in that she states directly that her aim

is to tackle the intermediate forms that can not be classified as belonging to one variety

or the other (p. 3). In the MLF model, she takes ʕāmijjə as the “matrix” (or dominant)

language and fuṣḥā as the “embedded” language. This is a striking theoretical statement

in that she is stepping away from the usual conception of fuṣḥā as the “standard”
 19

language that speakers “deviate” from by using markers of ʕāmijjə, and also recognizes

the social reality that ʕāmijjə is the mother-tongue of Arabic speakers.

To deal with the intermediate forms, Boussofara-Omar proposes three processes:

colloquialization of fuṣḥā, fuṣḥā-ization of colloquial, and hyper-fuṣḥā-ization of

colloquial (p. 109). I think the first two processes are helpful theoretically, but I would

also add that we could be more specific in identifying the structural level of the

language where the process is occurring. For example, if a Tunisian speaker pronounces

a fuṣḥā word like ‫ ﺃأﻳﻀﺿ ًﺎ‬/ʔaɪjɪḍan/ as [ʔaɪjɪð̣an], this would be colloquialization of fuṣḥā

on the phonological level. If an Arabic speaker uses the ʕāmijjə relative pronoun ‫ ﺍاﻟﻠﻲ‬ɪllī

instead of the fuṣḥā ‫ ﺍاﻟﺬﺫذﻱي‬allɛðī, this would be colloquialization of fuṣḥā on the lexical

level. Boussofara-Omar does not provide any clear examples of fuṣḥā-ization of

colloquial.

Although Boussofara-Omar addresses these issues theoretically and

methodologically, these insights are lost when she actually presents her data. For

example, let us look at an excerpt from her data. Note that these data represent her own

transliteration and translation, which differs slightly from mine.

li-ʔawaal marra fii taarriix il-blɛɛd haað̣i tkawnat

to-first time in history DEF-country this develop 3SG PERF
F TA/F TA/F TA/F TA TA TA
 20

xaṭaaba bi-l-luuɤa d-dɛɛrjə

rhetoric with-language DEF-colloquial
TA F TA

‘For the first time in the history of this country the colloquial has come to be used in

speeches’ (Boussofara-Omar, 1999, p. 93).

In this example, the first word /li-ʔawwal/ is marked as fuṣḥā (F) due to the

presence of the glottal stop, which is lost in TA (as well as the majority of Arabic

dialects, if not all). The second word /marra/ is categorized as both fuṣḥā and TA

because it exists in both varieties. However, the categorization of /il-blɛɛd/ as TA raises

a number of issues that illustrate the shortcomings of the code-switching framework.

Bourguiba has taken the fuṣḥā word ‫ ﺍاﻟﺒﺑﻠﺪﺩد‬/al-bɛləd/ and imposed TA phonology (saying

/il/ instead of /al/ for the definite marker) and syllable structure (CC instead of CV) on

the first part of the noun. This example would fall under what was described earlier as

“colloquialized fuṣḥā”, but this detail is lost in categorizing it as strictly TA due to the

phonology and syllable structure. Furthermore, this example raises the theoretical and

methodological question proposed earlier of whether such a slight change in phonology

can make a morpheme belong strictly to one variety and not the other.

This issue becomes even more pronounced by inconsistencies in coding in the

literature on code switching between different varieties of Arabic. First of all, among
 21

code-switching studies like that of Boussofara-Omar and Bassiouney (2006), there are

inconsistencies in whether they analyze a word as a whole unit or morpheme by

morpheme. For example, below is another example of Boussofara-Omar’s data.

u yuqtul kul man ya-tawassam fii-h

and kill 3 SG IMP all who 3SG-expectIMP in-him
TA TA TA/F F F TA

‘And he kills anyone whom he suspects of rivalry’ (Boussofara-Omar, 1999, p. 143).

In this second example, she has analyzed the second word /yuqtul/ as one word

instead of separating /yu-/ as the verb inflection for subject, as she does with the phrase

/ya-tawassam/ three morphemes later. It is easier to analyze /yuqtul/ as TA since

Bourgiba is using ʕāmijjə syllable structure. However, if we separate this phrase as two

separate morphemes /yu-qtul/, it becomes an intermediate form.

The /yu-/ subject marker is shared between both varieties, and /qtul/ becomes

colloquialized fuṣḥā due to the ʕāmijjə syllable structure. Whether or not we can classify

the morpheme /qtul/ as belonging to ʕāmijjə simply due to the syllable structure

depends on whether we think it is possible to classify this morpheme as belonging

strictly to ʕāmijjə by this slight modification.

This issue in fact represents another inconsistency in the coding of these studies,

which is inconsistency in labeling these phonological differences. For example, in fuṣḥā

 22

‫‘ ﻮﻭو‬and’ is pronounced as wə, whereas in many dialects it is common to pronounce this

morpheme as u. Bassiouney (2006) classifies wə as fuṣḥā and u as ʕāmijjə and

Boussofara-Omar follows suit. However, Bassiouney does not take the same approach

for verbs. For example, when the verb ‫ ﻛﺎﻦﻥن‬kānə ‘he/it was’ is pronounced as kān, she

marks this as a mixed form. In both examples one phoneme is lost, yet they are

classified differently.

I believe this is one of the main problems with the code-switching framework,

because in order to classify a morpheme as belonging to one variety and not the other

we make artificial, piecemeal, and inconsistent assumptions about the nature of

language, accent, and linguistic variation. Furthermore, such a methodology ignores

what Labov (1969) called the “inherent variability” of language, the fact that speakers

may use two different variants in the same conversation. Although the code-switching

framework tried to take a more fluid view of linguistic variation like Labov by

recognizing that speakers can switch between codes within the same utterance, its

reliance on binary classification remains inadequate for handling shared and

intermediate forms.
 23

Style Shifting: A New Approach

Finally, the last framework that has been used to describe linguistic variation is

style shifting. Instead of focusing on particular codes, this framework more broadly

conceptualizes speakers as alternating between various dynamic styles. While Mejdell

(2006) is perhaps the most extensive analysis of mixed styles of Arabic, she does not

take Labov’s variationist approach, which is used here. Mejdell includes levels of

Arabic, code switching, and ESA under a broad umbrella of ‘style’ that is largely vague

and undefined.

I find Allan Bell's (1984) theory of style shifting and audience design more

useful for the purposes of this study. Bell defined style as “essentially, a speaker’s

response to their audience” (p. 145). The audience consists of different groups of

individuals that play various roles in the conversation. The main interlocutor, and

therefore, the main influence on the speaker, that the speaker is holding a conversation

with is the addressee. Third persons who are present but not directly addressed are

auditors of the conversation; third parties whom the speaker knows are present but are

not ratified participants are overhearers of the conversation; and other parties whose

presence is unknown are eavesdroppers. Finally, other individuals that may exert

influence on a conversation are the referees, who are “third persons not physically
 24

present at an interaction, but possessing such salience for a speaker that they influence

speech even in their absence” (Bell, 1984, p. 186).

Bell criticized Labov’s (1972) claim that style can be solely measured by the

amount of “attention paid to speech”, which entailed that the more attention a speaker

pays to what they are saying, the more formal the style will be, for lacking empirical

evidence and insufficiently explaining the complexities of style. Bell proposed

conceiving of attention as “a mechanism, through which other factors affect style...

[that] is at most a mechanism of response intervening between a situation and a style”

(p. 150). Other factors like the topic of the conversation, the addressee, and setting can

also affect how much a speaker is paying attention to the way they speak.

Setting, more or less equivalent to domain, is an important factor in audience

design because it can determine the addressee or intended audience. For example, if a

person invites a friend over to their house for dinner, the intended audience is merely

the person with whom they are holding the conversation. In the workplace when a boss

is holding a company-wide meeting, the audience can be not only the workers present

at the meeting, but also referees like company shareholders that influence the

conversation despite the fact that they are not present. The conversation between

friends in a private home is a relatively informal domain, whereas the professional

 25

environment of a company meeting is a relatively formal domain, and these have

consequences for the type of language that is expected in each. In the former, it would

be perfectly acceptable to say “We’re gonna start soon”, whereas in the latter situation it

would probably be more appropriate to say “We’re going to start soon”.

In Bell’s theory, stylistic variation, which “denotes differences within the speech

of a single speaker” is placed on one of two axes of linguistic variation, the other being

the “social” dimension of variation, which “denotes differences between the speech of

different speakers” (p. 145). A linguistic analysis of stylistic variation, like this study,

investigates the speech of individuals in different styles (formal vs. informal) and

settings (in studio vs. on the street), whereas a linguistic analysis of social variation

would investigate the speech of individuals based on social factors like age, sex,

socioeconomic status, etc. Furthermore, stylistic variation occurs within the individual

(intraspeaker), whereas social variation occurs between different individuals

(interspeaker).

The audience design theory of style shifting is closely related to Giles and

Smith’s (1979) theory of speech accommodation, where speakers can “converge”

towards an interlocutor, or speak more similarly to the way they are speaking, or they

can “diverge” away from the interlocutors, socially distancing themselves. Speakers can
 26

use convergence to mitigate social distance, establish solidarity, or provide a friendly

atmosphere, whereas divergence can exaggerate social distance, and may be used in

order to establish authority or prove one speaker is more educated than the other. In

the context of style shifting, speakers can respond to their audience either through

convergence, by speaking in a similar style, or they can diverge by employing a

different style than what is expected by the interlocutor.

Conclusion

In this chapter, I have shown the weaknesses of the various theories that have

sought to explain linguistic variation in Arabic (diglossia, levels, code switching,

Educated Spoken Arabic) and described the theoretical framework taken here (the

audience design theory of style shifting). The present study will test the hypotheses

proposed by these theories through an examination of linguistic variables on the

phonological, morphological, and syntactic levels of the language.

The theory of diglossia would predict that the H variety will be used in formal

domains and the L variety will be used in informal domains. For this study, the formal

domain would be when the host and guests are in studio, whereas the informal domain

would be in the home or on the streets. Therefore, according to diglossia, when the host
 27

and guests are in studio they should use only H features while the host and guests on

the street should only use L features.

Badawi’s levels theory would predict that the in-studio Arabic speakers will be

placed closer to the fuṣḥā end of the continuum, perhaps around the fuṣḥā al-ʕaṣr level,

whereas the on-the-street speakers will be closer to the ʕāmijjə end of the continuum,

perhaps around the ʕāmijjət al-muθəqqəfīn or ʕāmijjət al-mutanawwɪrīn level. However,

since it is not clear what places a speaker at a particular level, this hypothesis will not

be tested in this study.

The code-switching framework would predict that the in-studio speakers

maintain more fuṣḥā code markers due to the formal domain, whereas the on-the-street

speakers will maintain more ʕāmijjə code markers due to the informality of the street

domain.

However, since code switching does not predict or explain intermediate forms it

has limited applicability. Furthermore, since the question of what to do with the shared

forms remains unanswered, so much data would have to be disregarded that the code-

switching framework would be difficult to apply. This issue will be explored later in the

analysis of the data, where I will attempt to show how pervasive shared and

intermediate forms are in actual speech.

 28

Finally, style shifting would predict that the in-studio speakers use a more

formal style compared to the speakers on the street. Although Bell criticized Labov’s

theory of attention paid to speech as inadequate, I think it does have some limited

applicability in this study, which will be discussed below. The style shifting theory will

be tested by examining two phonological variables (ḍ) and (ð̣), as well as two lexical

and syntactic variables: particles of negation and relative pronouns. Therefore, style

shifting would predict that the in-studio guests use a more formal style by maintaining

the fuṣḥā [ḍ] variant whereas the on-the-street guests will use more of the ʕāmijjə [ð̣]

variant. For the variable (ð̣), the Egyptians are expected to use more of their colloquial

variant [ẓ] than the other guests. Finally, the in-studio guests are expected to use more

fuṣḥā particles of negation and relative pronouns compared to those guests on the street.

I will show that the style-shifting framework is best suited for the nature of this

study because unlike the other theories discussed in this chapter it allows for analysis of

shared and intermediate forms.

In the next chapter, I will discuss the Al-Jazeera television program ḥuwār

mɛftūḥ, where the data for this sociolinguistic study was obtained, as well as the

phonological, morphological, and syntactic variables for this study.

 CHAPTER 3

THE PROGRAM AND LINGUISTIC VARIABLES

The Media Context

The broadcast media context has linguistic consequences of the type of language

that is expected. Arabic speakers on Al-Jazeera know they are being broadcast across a

vast area that covers an array of dialects. Therefore, they are likely to use fuṣḥā so that

ideally they will be understood by as many Arabic speakers as possible. The place of

fuṣḥā in the media context is a somewhat idealized lingua franca that facilitates

interdialectal communication.

However, it must be noted that fuṣḥā can only be considered a lingua franca

among educated Arabs. Proficiency in the standard language is achieved by native

speakers through intense education, and most Arabic speakers do not receive advanced

training in the standard unless they are studying to become Arabic teachers or linguists.

Even though some speakers receive advanced instruction in the standard, certain

features or rules may be lost with lack of use over time and not being exposed to the

standard variety. Although Al-Jazeera is highly popular and can usually be seen on TVs
 30

in cafes or shops on a daily basis throughout the Middle East, the type of audience that

tunes into the news is assumed to be somewhat educated in the first place, which is why

they are keeping up with current events. However, the type of program can also entail

how educated the audience is that is tuning in. For instance, the popular roundtable

discussion program on Al-Jazeera like ‫‘ ﺍاﻻﺗﺠﺟﺎﻩه ﺍاﻟﻤﻣﻌﺎﻛﺲﺱس‬The Opposite Direction’, hosted by

Faisal Al-Qassim, might not necessarily attract a particularly educated audience because

of its tendency to sensationalize confrontation. Viewers may tune in to a show like ‘The

Opposite Direction’ just to see two individuals with drastically different viewpoints

duke it out, with little focus on the substance of the debate. I think that with ‘Open

Dialogue’, the program under investigation here, since there is more focus on

substantive debate and less sensationalizing, we could assume that the type of audience

for this program may be more educated than the casual viewer just looking for the

headlines or heated confrontation.

An educated audience can also entail that these viewers are familiar with

regional differences, whether phonological or lexical, and may be comfortable with

some level of dialectal markers on Al-Jazeera, otherwise we could assume that they

would change the channel if they do not understand the conversation. Therefore, the

media context provides certain expectations about the type of audience the guests on
 31

the program think they are addressing and the kind of linguistic markers they will use.

All of this makes media broadcasts an ideal testing ground for the various theories being

examined.

The Program

Data for this study was taken from the Al-Jazeera program ‫ ﻤﻣﻔﺘﻮﻭوﺡح ﺤﺣﻮﻭوﺍاﺭر‬ḥuwār

mɛftūḥ, ‘Open Dialogue’, which is no longer on the air. Ghassan Bin Jiddu, who is part

Tunisian and part Lebanese hosted the program. According to his Facebook profile

(https://www.facebook.com/Ghassan.ben.jiddo, accessed February 6, 2012), he went to

college in Tunisia and then moved to Lebanon to work as a reporter for Al-Jazeera. He

resigned his post as head of the Al-Jazeera station in Beirut in April 2011 to protest the

fact that the Qatar-based station did not give adequate coverage to the uprisings in

Bahrain and Syria.

This program was chosen mainly because the Arabic scripts were available on

the Al-Jazeera website (http://www.aljazeera.net/programs/27d796cb-abce-444f-a271-

24c83bfc051c, accessed December 15, 2011). The videos were downloaded in iTunes

through the podcast of the program, and the MPEG-4 video files were converted to .wav

format using Smart Converter to allow for phonological analysis where necessary. The
 32

program is an interesting setting to analyze interdialectal conversation since many of

the guests the host interviews come from various countries and social backgrounds.

Furthermore, the setup of the program offers several speaking styles. At the beginning

of each program, Bin Jiddu reads from notecards giving an introduction to the

conversation topic for the program, and at the end he reads a farewell message

thanking the viewers for tuning in and those involved in producing the program. The

panel discussions provide a fertile ground for voicing one’s opinion, arguing,

summarizing or paraphrasing what other people have said, joking, and debating.

For the most part the panelists address the host and not each other, so the setup

of the conversations is predominantly one-on-one. Therefore, in the audience design

framework the host and a guest will switch between roles of speaker and addressee.

While the host interacts with a single guest, the other guests can be considered auditors

to the conversation because the two main speakers know that they are there but do not

address them. Finally, the pan-Arab audience that is watching the program can be

considered the referee, because even though they are not present in the conversation,

they still exert an influence because the program is being broadcast to them, so it is

upon all participants to seek to be understood by them.

 33

The Episodes

The first episode titled ‘Intellectuals and the Media in the Time of Revolution,’

aired on Al-Jazeera on February 20, 2011, nine days after Hosni Mubarak stepped down

as president of Egypt. The discussion panel includes Khalid Yusef, a director, Nuwara

Negm, a female blogger, and Gamal Bakhit, a poet. Khalid Yusef is famous in Egypt for

his blunt, controversial films that handle taboo topics like rape, homosexuality, and

political corruption. Nuwara Negm became somewhat of a media spokesperson for the

protesters during the revolution when she was interviewed by Al-Jazeera on January

26, 2011. Hosni Mubarak had just made a speech dismissing the protests as a temporary

phenomenon, and in the interview, she bluntly stated that the protesters weren’t going

anywhere until Mubarak stepped down, which solidified the determination and

persistence of the protesters in Tahrir square. A translator and news editor at the Nile

Television Network, she writes about political issues at her blog, ‫ ﺍاﻟﺸﺷﻌﺒﺑﻴﺔ ﺍاﻟﺘﻬﮭﮫﻴﻴﺲﺱس ﺠﺟﺒﺑﻬﮭﮫﺔ‬, gɛbhɛt

ɛt-təhīs ɛ-šaʕbijə (http://tahyyes.blogspot.com/, accessed February 6, 2012), which I will

translate as ‘A Popular Front of Sarcasm’. Gamal Bakhit is a renowned Egyptian poet,

whose poem ‫ ﺭرﺍاﺴﺳﻚﻙك ﺍاﺭرﻓﻊ‬،٬‫ﻤﻣﺼﺻﺮﻱي ﺍاﻨﻧﺖﺕت ﻓﻮﻭو‬, ɪrfaʕ rasɪk fu, ɪntə məsrī, which translates as ‘Raise

Your Head High, You Are Egyptian’, captured the protester’s demands for a government

that respected their dignity as human beings (Khawly, 2012).

 34

The second episode titled ‘Developments on the Situation in Tunisia’ aired

January 15, 2011, one day after the Tunisian president Zine Al-Abadine Ben ‘Ali

stepped down and fled to Saudi Arabia. The in-studio discussion panel includes Rashid

Al-Ghannouchi, one of the leaders of the ‫ ﺍاﻟﻨﻧﻬﮭﮫﻀﺿﺔ‬ɛn-naḥḍə political party, which won the

majority of seats in the first free Parliamentary elections in Tunisia’s history in October

of 2011, as well as the Palestinian editor of the London-based pan-Arab newspaper Al-

Quds, ‘Abdul Bari Al-‘Atwan. Siham Bensedrine, a Tunisian human rights activist and

journalist, takes part in the conversation over the phone, and Jamal Khashoggi, a Saudi

journalist, participates via videoconference.

The third episode, which is simply titled ‘Tunisia’ aired on January 22, 2011 and

consists of three parts. In the first section, the host Ghassan Bin Jiddu is out in the

streets of Tunisia interviewing attendees at a local rally. In the second section, he hosts

a panel discussion with two Tunisian journalists, Lutfi Haji and Ziad Tarbush. In the

third section, he interviews a Moroccan journalist for Al-Jazeera, Anas Bin Saleh, who

was stationed in Tunisia to cover the revolution.

 35

Variables

Phonological

For this study, the choice of the phonological variable ‫( ﺾﺽض‬ḍ) was primarily

dictated by the data, because in listening to the episodes closely variation in this

phoneme seemed the most salient. In the Tunisian dialect as well as most dialects, it is

extremely common for the variable ‫( ﺾﺽض‬ḍ) to be pronounced as ‫[ ﻈﻇﻆﻅظ‬ð̣]. Watching Al-

Jazeera, one can hear this variation between [ḍ] as [ð̣] in every dialect except Egyptian,

because in the Egyptian dialect (ḍ) is sometimes pronounced as [ẓ].

Although no explicit sociolinguistic studies on the [ð̣] variant of (ḍ) were found

in the literature for any dialect, Ferguson (1959) mentions the merger of /ḍ/ and [ð̣] as

a feature of the Arabic koine except in dialects that have lost the interdentals (like

Egyptian) (p. 67). In addition, Talmoudi's (1981) description of the Tunisian dialect of

Soussa, Amor's (1990) Beginner's Course in Tunisian Arabic for the Peace Corps, and

Boussofara-Omar's (1999) dissertation on Tunisian Arabic use the same phonetic symbol

/ð̣/ for the Arabic letters ‫ ﺾﺽض‬and ‫ﻈﻇﻆﻅظ‬, suggesting that these have merged in TA.

Perhaps the lack of sociolinguistic investigation of this variable could be due to

the fact that researchers interested in the emphatics are generally more concerned with

phonological processes associated with the voice quality rather than place of
 36

articulation. Most studies on the emphatics in Arabic are primarily phonological studies

of emphasis spread (Watson, 2002) or phonetic characteristics of emphatics

(Abudalbuh, 2010; Al-Masri, 1998; Boxberger 1981; Muqbil, 2006). Unfortunately,

many of these studies are phonetic experiments where the subjects read a word list,

which means they may be consciously trying to use the fuṣḥā pronunciation making

them less useful for comparison for linguistic variation.

The second phonological variable in this study is ‫( ﻈﻇﻆﻅظ‬ð̣) > [ẓ] in Egyptian, so that

a fuṣḥā word like ‫ ﻤﻣﻈﻇﻆﻅظﺎﻫﮬﮪھﺮﺍاﺖﺕت‬/muð̣āhɪrāt/ ‘protests; demonstrations’ may be pronounced as

/muẓāhɪrāt/ in Egyptian. Watson (2002) notes that Cairene Arabic has lost the

interdental fricatives (p. 20), and Soliman (2008) the merger of /ð̣/ > [ẓ] (p. 84).

These findings follow the trend of other dialects where interdentals are pronounced as

alveolars. For example, in many other dialects (Levantine, Gulf, Magrebi) it is extremely

common to pronounce ‫ ﺬﺫذﻫﮬﮪھﺐﺏب‬/ðəhəbə/ 'to go', as ‫ ﺯزﻫﮬﮪھﺐﺏب‬/zəhəbə/. Therefore, it is not

surprising that Egyptians avoid pronouncing (ḍ) as [ð̣] because interdental fricatives are

hardly used in the dialect in the first place. Because of this avoidance of interdentals,

one could speculate that the evolution of the sound change /ḍ/ > [ð̣] > [ẓ] would

make sense in Egyptian.

 37

Lexical and Syntactic

Two other categories of variables were chosen for this study: particles of

negation and relative pronouns. These variables were chosen because they were

particularly salient in the data set, allowing several tokens to be collected from each

speaker. Particles of negation and relative pronouns will illustrate that style shifting

does not occur solely on the phonological level of the language, but the lexical and

syntactic level as well. It should be noted that the negation particles and relative

pronouns are considered lexical variables because there is a choice between one word

and the other, and syntactic variables since these particles work differently in the

grammar of both varieties.

Negation

In fuṣḥā there is one particle for nominal negation ‫ ﻟﻴﺲﺱس‬lejsə, and four particles for

verbal negation that depend on the tense of the verb: ‫ ﻟﻢﻡم‬lɛm and ‫ ﻤﻣﺎ‬mā for the past, ‫ ﻻ‬lā

for the present, and ‫ ﻟﻦﻥن‬lɛn for the future, whereas in the dialects ‫ ﻤﻣﺶﺵش‬mɪš/muš or the

circumfix ‫ ﻤﻣﺎ‬- ‫ ﺶﺵش‬ma-š are used.

The negation particle lā serves several functions outside of negating the present

tense in both fuṣḥā and ʕāmijjə. First, it can be used simply to mean 'no', as in response

to a question. Second, lā can also be used to indicate categorical negation, as in ‫ﻻ ﺸﺷﻚﻙك‬

 38

lā šɛk, meaning 'no doubt'. Third, lā can be used for the negative imperative, and

finally, when it is repeated it can indicate ‘neither…nor’. For this study, when lā was

used in these other functions they were ignored, and only instances where there was a

clear choice between a fuṣḥā or ʕāmijjə particle of negation were used.

Relative Pronouns

In fuṣḥā there are three relative pronouns depending on gender and number of

the noun being referred to: ‫ ﺍاﻟﺬﺫذﻱي‬allɛðī for masculine singular, ‫ ﺍاﻟﺘﻲ‬allɛtī for feminine

singular, and َ‫ ﺍاﻟﺬﺫذﻳﻦﻥن‬allɛðīnə for plural, whereas in the dialects only ‫ ﺍاﻟﻠﻲ‬ɪllī is used. Again,

some linguists may classify the use of ɪllī as a diglossic choice because it does not exist

in fuṣḥā.

Interestingly, Walters (2003) claims that in what he refers to as Oral Literary

Arabic, which is similar to the concept of Educated Spoken Arabic, it has become

“highly conventionalized” for Tunisians to use the dialectal rather than the fuṣḥā

relative pronoun (p. 101). This present study will test this hypothesis.

Other Variables

I have chosen two phonological variables (ḍ and ð̣), as well as two lexical and

syntactic variables (particles of negation and relative pronouns) in order to offer a

 39

glimpse of style shifting in Arabic. After the results are presented for these variables, I

will discuss other variables such as the use of case endings or certain inflectional

markers where the speakers in this data set exhibited variation in using fuṣḥā or ʕāmijjə

markers. This analysis will illustrate the shortcomings of the diglossic and code-

switching frameworks, as well as provide a more holistic view of stylistic variation in

Arabic.

Conclusion

In this chapter, I have introduced the broadcast program and variables under

review. In the next chapter, I will discuss the methodology of this study, detailing how

the data were collected and analyzed.

 CHAPTER 4

METHODOLOGY

Data Collection

In this section, I describe how tokens of the phonological, lexical, and syntactic

variables were gathered. As stated earlier, one of the main reasons this specific program

was chosen was because the Arabic transcripts were available on the Al-Jazeera

website. First, the Arabic transcripts were transliterated into Latin orthography to ease

the process of phonetic transcriptions of each token. Once the transliteration was

complete, words containing the phonemes (ḍ) and (ð̣) were identified in the transcripts

to make it easier to follow along with the audio.

To obtain the sound files of the individual tokens, Audacity software was used to

select sections of the wavelength of the sound file where the tokens occurred. These

selections were then exported as individual sound files that were coded by speaker and

number. These tokens were entered into a spreadsheet where they were phonetically

transcribed coding for speaker, dialect group, and setting.

 41

Figure 1 shows an example of how the variables were organized in a

spreadsheet.

The setup depicted in Figure 1 allowed close inspection of tokens that were not

clear. Using headphones to remove background noise, each token was listened to closely

several times. If a token was unclear, its sound file was analyzed in Praat to determine

which variant was being used.

For example, Figure 2 shows how the wavelengths of the two phonemes differ.

Figure 1. Token spreadsheet.

Figure 2. (ḍ) and (ð̣) spectrogram.

 42

In Figure 2, the phoneme on the left represents Ghassan’s pronunciation of /ḍ/ in the

word ‫ ﺒﺑﻌﺾﺽض‬baʕḍ ‘some’. The tall and spacious wavelengths suggest that this phoneme is a

stop, meaning he is using the [ḍ] and not the fricative [ð̣]. The phoneme on the right

represents Ghassan’s pronunciation of the /ḍ/ in the word ‫ ﺭرﻓﺾﺽض‬rəfḍ ‘he refused’. Here

the wavelengths are short and bunched together, showing the frication or noise in the

pronunciation of [ð̣], the fricative, and his dialectal variant. The analysis of the sound

files was conducted in this manner to determine which variant was being used.

Some tokens had to be eliminated because of unclear audio, overlapping speech,

background noise, and the rapidity of utterances. One Tunisian speaker on the street

was eliminated because he became so emotional describing his experiences of torture at

the hands of the Tunisian authorities that he could not be understood.

For the lexical and syntactic variables, the phonetic transcripts of each episode

were examined. Each particle of negation and relative pronoun was counted and

categorized as a fuṣḥə or ʕāmijjə lexical item and tallies were calculated for each

speaker.
 43

Categorization of Speakers: Speech

Setting and Interlocutor

Because this study analyzes style shifting through the lens of audience design,

the question of whom the host and guests are addressing is of utmost importance in

how they choose which style to employ. Both the host and guests can be said to be

addressing the pan-Arab audience (the referee in Bell's framework) watching the

program, which could perhaps entail more use of features from fuṣḥā, the supposed

lingua franca of the educated viewing audience as discussed previously. The guests

primarily address the host. However, as the moderator of the discussion, the host

switches between addressing Egyptians, in-studio Tunisians, on-the-street Tunisians,

interdialectal panels, and the Tunisian woman on the phone.

Setting is another important element of style shifting because it may help to

emphasize the intended audience. Between the three episodes, the host was either in-

studio with a homogenous dialect group (the Egyptians), in-studio with speakers of

various dialects, or out in the streets interviewing Tunisians. The guests ranged from a

homogenous in-studio panel (the Egyptians in one episode and the two Tunisian

journalists in another), an interdialectal panel (Rachid Al-Ghannouchi and ‘Abdul Bari

Al-‘Atwan, a Tunisian and a Palestinian, as well as Ghassan's discussion with Anis Bin
 44

Saleh, a Moroccan), over the phone (Siham Bensedrine, a Tunisian), videoconference

(Jamal Khashoggi, a Saudi), and Tunisians on the street. Therefore, the guests were

categorized into groups depending on the setting of the conversation, which is

illustrated in Table 1.

It should be noted that these categorizations depicted in Table 1 may be

somewhat incomplete in the sense that speakers are not solely defined by the setting in

which they are speaking. Such a categorization is by no means deterministic, because

linguistic variation cannot be explained by one factor alone. Here, they provide a

starting point for comparison.

Table 1. Categorization of Guests

In-Studio In-Studio Various Dialects On-the-Street Tunisian Woman

Egyptians Tunisians Tunisians On-the-Phone

Khalid Rachid ‘Abdul Bari Al- 11 total Siham Bensedrine

Yusef Al- ‘Atwan (Palestinian)
Ghannouchi

Nuwara Lutfi Haji Jamal Khashoggi

Negm (Saudi)

Gamal Ziad Tarbush Anis Bin Saleh

Bakit (Moroccan)
 45

The setting also determines how aware the speakers are of their language, so

here Labov’s ‘attention paid to speech’ can be useful as well. In studio, it is assumed

that the speakers will be quite aware of the formality of the situation and the language

required. However, as the show progresses and the participants get deeper and deeper

into conversation, there is a possibility that the speakers will become less aware of the

studio setting and may start using more dialectal features. Walters (2003) writes that

“even in fairly formal situations like radio and television interviews, the longer the

interlocutors interact, the more likely they are to ‘drift’ toward the dialect” (p. 92). At

the beginning of the episode, both the host and guests may be quite aware of the pan-

Arab audience watching at home and cater their speech to that audience. However, in

the middle of the episode a guest may be so drawn into the conversation that they are

focusing more on the host as their immediate audience than those tuning in.

This shift in awareness of audience is also related to how often the guests are

permitted to speak. The in-studio Egyptians, for instance, are the only three guests on

the program. They therefore have more opportunities to speak compared to the guests

in other episodes. The larger amount of airtime could also allow the Egyptians to

become less aware of the studio setting as they are caught up in the conversation. It

would be interesting to track the variation over the course of the episode; however, the
 46

setup of the program prevents such an investigation since the host addresses all three

guests throughout the program at various intervals. For example, if the host talks to

Khalid Yusef for the first 10 minutes, and then talks to Nuwara Negm for 10 minutes, by

the time he gets back to Khalid Yusef it is hard to say how aware of his setting and

audience he is. Furthermore, such an investigation would require looking at more

variables on the phonological, lexical, and syntactic levels because I do not have enough

data to perform such an investigation. However, future research could investigate this

issue because it may offer insight into conscious awareness of setting and audience.

The on-the-street Tunisians were usually interviewed very briefly and therefore

had less time to speak compared to the Egyptians. Additionally, although they were on

the street, which according to the diglossic framework is a more informal domain, they

have a camera in their face and they are being interviewed by a reporter, which may

make them more aware of their speech. They may pay more attention to their speech

because they know they are on Al-Jazeera and whatever they say is being broadcast

across the Arabic-speaking world. Furthermore, since the on-the-street Tunisians were

only interviewed for a short time they have less time to forget about the presence of the

camera.
 47

The in-studio Tunisians, on the other hand, we can assume are highly educated

because they are considered authoritative figures on the topics being discussed, whereas

the average Tunisian on-the-street may not be as educated, meaning that they may use

more dialectal features simply because they have not received as much education in

fuṣḥā. The Tunisian woman on-the-phone adds another interesting element to this issue,

because it is my belief that she may not be as aware of the presence of the cameras

because she is speaking from a phone in her home. The informal setting of being in the

comfort of her own home and the fact that a camera is not in front of her face may lead

her to speak more naturally, which may entail more ʕāmijjə features.

Finally, another result of the guests’ being allotted different time to speak has

consequences on the number of tokens that can be obtained from each of the speakers

in this study, which has implications for the comparability of the results. The greatest

number of tokens for each variable was obtained from the speech of the host because he

usually talked the most and was present in all three episodes. However, these tokens

were dispersed across a variety of styles: reading, addressing various guests (Egyptians,

Tunisians, speakers from other dialects), and addressing the general audience (whether

in studio or in the street). So for example, since the host only reads at the beginning of
 48

the program, there were very few tokens obtained in the reading style compared to

others.

The number of guests on the program also determines how long the host has

time to speak. For the on-the-street interviews, the host spoke with 11 Tunisians total.

The interviews were quite short for the most part, except for three men who spoke at

length. Since the name of this segment of the show was ‘Unheard Voices,’ the point was

to get the Tunisian perspective and not the host's. Therefore, only eight tokens were

obtained from the host in this context. Since the Tunisians spoke briefly in this context,

all tokens were combined into a group labeled “On-the-Street Tunisians.” The other in-

studio guests had more time to speak than the on-the-street interviews; therefore

around twenty tokens were obtained from most of the in-studio guests.
 CHAPTER 5

RESULTS AND DISCUSSION

Phonological Variable: ‫( ﺾﺽض‬ḍ) > ‫ ﻈﻇﻆﻅظ‬/ð̣/

The Host

Table 2 summarizes the results for phonological variation of (ḍ) and [ð̣] in the

speech of the host of the program, Ghassan Bin Jiddu, depending on whom he is

addressing.

In order to analyze these results, percentages were calculated for each variant,

and the results are represented in Figure 3.

In Figure 3, we can see that the host has the highest percentage of the ʕāmijjə

variant [ð̣] in his speech phonologically when he is talking to the Egyptians. We can

explain this as speech accommodation to some extent but not completely. Usually with

speech accommodation, language users will speak more similarly to the speech of their

interlocutor.
 50

Table 2. Counts of tokens of (ḍ) produced by host

ḍ % ð̣ % Total

Egyptians 1 4% 27 96% 28

Tunisian woman on-the-phone 1 9% 10 91% 11

In-studio Tunisians 3 10% 27 90% 30

Various dialects 4 27% 11 73% 15

On-the-street Tunisians 3 33% 6 67% 9

General audience (street) 5 42% 7 58% 12

Reading 18 47% 20 53% 38

General audience (studio) 4 27% 11 73% 15

Total 39 25% 119 75% 158

 51

Host (ḍ) > [ð̣] and [ḍ]

100%
90%
80%
70% ḍ
60%
50% ð̣
40%
30%
20%
10%
0%

Figure 3. Host (ḍ) variation

However, we cannot necessarily say that Ghassan Bin Jiddu is accommodating

to the Egyptians by speaking the Egyptian dialect because (ḍ) > [ð̣] is not a feature of

the Egyptian dialect. For Egyptians, the change is towards [ẓ] instead of [ð̣], although it

is not clear how common this variant is. I think it is more preferable to say that he is

accommodating to their colloquial style of speaking rather than to say that he is

accommodating to their colloquial variety. As will be discussed below, the Egyptians

speak in a predominantly colloquial style in several different ways.

 52

After the Egyptians, the Tunisian woman on the phone (the journalist and

human rights activist Siham Bensedrine) is the next person with whom he

predominantly uses his dialectal [ð̣] variant. Similarly, I think this is related to the fact

that she speaks in a more colloquial style when compared to her male Tunisian

counterparts. This could be related to the fact that women in general are said to used

more colloquial features in general (Al-Wer 1999; Bakir, 1986; Daher, 1998, 1999), as

well as to the fact that she is on the phone in her own home. Furthermore, her use of

colloquial features may be connected to her role as a populist activist, reflecting her

role as an advocate of the people.

Surprisingly, Ghassan Bin Jiddu uses his dialectal variant more with the in-

studio Tunisians as compared to those on the streets. This could be a comparability

issue related to the fact that while 30 tokens for this variable were captured in studio,

only 9 were obtained in the on-the-street interviews. This is related to the lower quality

of the on-the-street recording where an abundance of background noise on the street

resulted in several tokens being thrown out. He also did not speak very much on the

street compared to other settings. It is possible that if more tokens were obtained from

the street setting the results would be reversed for these two groups he is addressing.
 53

The results for this variable are also interesting when the host addresses the

general audience on the street and in studio. The fact that he uses his dialectal variant

more on the street than in studio is in line with my prediction that he would use more

ʕāmijjə markers in general on the street. These findings are particularly interesting

regarding the general audience because as we discussed earlier, since Al-Jazeera is

broadcast across the Arab world it has to be accessible to audiences from various

dialectical backgrounds. This phonological variation does not necessarily prevent

comprehension among these diverse groups since this variation between [ḍ] and [ð̣]

may be present in every dialect except Egyptian. Such a phenomenon goes back to the

question of whether Al-Jazeera needs to be broadcast in fuṣḥā, the supposed “mutually

understood standard”, or if speakers on Al-Jazeera can use fuṣḥā with some widespread

phonological ʕāmijjə markers, which may constitute a third or intermediate variety that

is often hypothesized (ESA).

The results for Ghassan Bin Jiddu's phonological variation in the reading context

are particularly interesting because it is nearly half and half. According to Labov's idea

of “attention paid to speech”, reading is supposed to be the most formal setting where

we are most aware of the language we are producing. The fact that the host uses nearly

50% fuṣḥā pronunciation and 50% ʕāmijjə pronunciation in this context can have
 54

several interpretations. Ghassan Bin Jiddu could be aware of this phonological variation

in his dialect and he is trying his best to control it. On the other hand, he could be

unaware of this difference in his dialect and switches between the two outside the realm

of conscious awareness. Although we cannot know for sure either way, these results

draw an interesting nuance in the relationship between attention paid to speech and

conversational setting or type.

Finally, it is not surprising that he uses the most fuṣḥā pronunciations of this

variable when he is addressing the general audience in the studio setting, again playing

his role of authoritative moderator and representative of the station.

Perhaps most importantly, the results for this phonological variable (ḍ) show

that the host did not maintain the fuṣḥā pronunciation 100% of the time in the formal

domain. His variation between [ḍ] and [ð̣] violated the functional differentiation

between fuṣḥā and ʕāmijjə proposed in the diglossic framework, and illustrated the

leakage between standard and nonstandard varieties. These results also exemplify

Labov’s notion of “inherent variability” since sometimes the host would say [ʔaɪjɪḍan]

‘also’ and other times he would say [ʔaɪjɪð̣an] depending on his audience. Style shifting

with this phoneme therefore allowed him not only to accommodate to individuals who

shared the same dialectal feature like the on-the-street Tunisians, but also permitted
 55

him to accommodate to the overall colloquial style in which the Egyptians were

speaking.

The Guests

Table 3 summarizes the results for phonological variation for the variable (ḍ) >

[ð̣] and [ẓ] for the guests on the program.

Percentages were calculated for the variants and are illustrated in Figure 4.

Figure 4 shows that the results for the guests’ phonological variation between [ḍ], [ð̣],

and [ẓ]. The on-the-street Tunisians used the ʕāmijjə [ð̣] variant 100% of the time. This

result is not surprising in the context of the audience design theory of style shifting

given the informality of the street domain, as well as their immediate audience, a fellow

Tunisian. The in-studio Tunisians used this variant the majority of the time as well,

which was surprising given the formality of the domain. On the one hand, this ʕāmijjə

variant could be so ingrained in the dialect that it is used even when Tunisians are

speaking fuṣḥā, similar to how Egyptians can maintain the ʕāmijjə [g] variant of /ʤ/

when they are speaking fuṣḥā. On the other hand, we can also interpret the presence of

the [ḍ] variant in studio reflecting the formality implied by the in-studio domain and

perhaps the in-studio Tunisians’ awareness of it. We could also hypothesize that both
 56

groups of Tunisians were accommodating to the host’s dialect because he shares this

variation between [ḍ] and [ð̣], manipulating their speech according to their audience.

The results of the Tunisian woman on the phone, Siham Bensedrine’s, variation

on this phoneme are particularly mixed and interesting. While she predominantly

pronounces /ḍ/ as [ð̣], she uses the [ẓ] variant nearly as much. This is surprising since

it was noted according to the literature earlier that variation between [ḍ] and [ẓ] is

usually a feature of Egyptian and not Tunisian. Siham Bensedrine grew up in La Marsa,

a suburb of Tunis. Perhaps this variation between [ḍ] and [ẓ] is an undocumented

feature of the area, but it is difficult to know without more speakers. For the purposes

of this study, we can only speculate that she is using more dialectal variants because she

is talking on the phone in her home, a natural comfortable domain where she may be

Table 3. Counts of tokens of (ḍ) produced by guests

ḍ % ð̣ % ẓ % Total

Tunisians on-the-street 0 0% 58 100% 0 0% 58

Tunisians in-studio 6 16% 32 84% 0 0% 38

Tunisian woman on-the-phone 3 14% 10 45% 9 41% 22

Various dialects 43 98% 1 2% 0 0% 44

Egyptians in-studio 68 97% 0 0% 2 3% 70

Total 120 52% 100 43% 11 5% 232

 57

Guests (ḍ) > [ð̣] or [ẓ]

100%
90%
80%
ḍ
70%
60% ð̣
50%
40% ẓ
30%
20%
10%
0%
Tunisians On- Tunisians In- Tunisian Various Egyptians In-
the Street (58) Studio (38) Woman On- Dialects (44) Studio (70)
the-Phone
(22)

Figure 4. Guests (ḍ) variation.

less aware of the media context because she is not sitting in front of a camera.

Interestingly, she tended to pronounce other interdentals as fricatives, so this variation

could be a continuation of that trend.

With all of the Tunisians, it is also important to remember that these episodes

aired a few days after Ben ‘Ali resigned as president. The persistent use of the [ð̣]

variant could also signal nationalistic pride in reclaiming Tunisian identity. The on-the-

street Tunisians who were rallying in the streets embodied this victory physically and

linguistically.
 58

The various dialects group, which again consists of the Palestinian, Saudi, and

Moroccan speakers, showed much less variation with this variable, maintaining the

fuṣḥā [ḍ] pronunciation nearly 100% of the time. This is a little surprising since (ḍ) >

[ð̣] is a feature of all three dialects; however, we could say that they may have

suppressed this feature because of the formal nature of the studio domain. The only

variation for this group with this variable came from Jamal Khashoggi, the Saudi

journalist, who maintained the [ḍ] phoneme in eighteen lexical items, whereas he used

the [ð̣] variant only once. It should be noted that in that episode he talked the least, for

about 8 minutes. Perhaps if he had talked more this variant might have been more

obvious, and it would be interesting to find other interviews to see if this variant is

more apparent in situations where he is given more allotted time to speak.

The data for the Egyptians support the idea that variation between [ḍ] and [ẓ] is

rather rare and restricted in the speech of the Egyptian guests. Nuwara Negm used the

dialectal variant only once when she said [təhərīẓi] for ‘inciting, provocative’ instead of

/təhərīḍi/, and Khalid Yusef did as well when he said [muharɪẓan] for ‘inciter,

instigator’ instead of /muharɪḍan/. Interestingly, these are both from the same root,

which may be related to why there is variation in both pronunciations despite the fact

that they are from two different speakers. It should also be noted that Nuwara Negm
 59

uses this root at four different points in the conversation, including the exact same verb

/tə-harīḍī/, and there she maintained the [ḍ] pronunciation. Again, this variation

exemplifies Labov’s “inherent variability” because she used different pronunciations for

the exact same verb.

From this chart, we have seen that variation for the variable (ḍ) > [ð̣] and [ẓ] is

highly complex in the dialects of the guests on this program. However, this variable is

just one way that these speakers shift between different styles. Next, I will look at ‫( ﻈﻇﻆﻅظ‬ð̣)

> [ẓ] in the Egyptian dialect to show how Egyptians maintain a colloquial style with

this next variable. After that, I will turn to lexical and syntactic analysis of negation and

relative pronouns in order to add another dimension to style shifting for the other

dialect groups.

Phonological Variable: ‫( ﻈﻇﻆﻅظ‬ð̣) > [ẓ]

For the second section of phonological analysis for this study, I investigated the

Egyptian variable ‫( ﻈﻇﻆﻅظ‬ð̣) > [ẓ] in the data set in order to show another way guests on

the program manipulated their speech on the phonological level of the language.

Tokens for this variable were also collected for the host, Tunisians, and various dialect

groups, but all three groups maintained the /ð̣/ pronunciation nearly 100% of the time,

regardless of setting.
 60

Results for this variable are illustrated in Table 4.

Since the Egyptians showed the most variation on this variable, percentages

were calculated for each speaker and are illustrated in Figure 5.

In Figure 5, we can see that the Egyptians predominantly used the ʕāmijjə [ẓ]

pronunciation instead of the fuṣḥā /ð̣/ pronunciation, though Khalid Yusef and Gamal

Bakhit did use the fuṣḥā pronunciation one time each. These results add another

dimension of style shifting on the phonological level in the speech of the Egyptians. For

the first variable (ḍ), the Egyptians did not use their dialectal variant [ẓ] nearly as much

as I expected. However, for the variable (ð̣) we see a much larger use of the ʕāmijjə [ẓ]

variant. This illustrates how the Egyptians tended to use more ʕāmijjə markers in

Table 4. Counts of tokens of (ð̣) produced by host and guests

ð̣ % ẓ % Total

The host 52 100% 0 0% 52

In-studio Tunisians 31 100% 0 0% 31

On-the-street Tunisians 16 100% 0 0% 16

Tunisian woman on-the- 14 100% 0 0% 14

phone

Egyptians 2 6% 32 94% 34

Various dialects 29 97% 1 3% 30

Total 144 81% 33 19% 177

 61

Egyptians (ð̣) → [ẓ]

100%
90%
80%
70%
60% ð̣
50%
40% ẓ
30%
20%
10%
0%
Gamal Bakhit Khalid Yusef Nuwara Negm
(14) (12) (8)

Figure 5. Egyptians (ð̣) variation.

general, as well as a continuance of the trend of pronouncing interdentals as fricatives.

The fact that Nuwara Negm maintained the ʕāmijjə [ẓ] variant is also in line

with the previous results for Siham Bensedrine, in the sense that they used more ʕāmijjə

phonological markers than their male counterparts. This is in line with other research

on Arabic that has shown women tend to use more colloquial features than men (Al-

Wer, 1999 in Jordan; Bakir, 1986 in Iraq; Daher, 1998, 1999 in Syria; Havelova, 2000

in Palestine).

I mentioned previously that the Egyptian dialect is widely understood in the

Arab world and one hears markers of the Egyptian dialect on several Arab media

outlets. Therefore, the Egyptians are not necessarily accommodating to the speech of
 62

the host because he does not vary on this variable. Rather, they are conveying their

Egyptian identity to the pan-Arab audience watching at home. Remember that this

episode aired a few days after Mubarak resigned as president; therefore, they could

have been using predominantly Egyptian colloquial features in order to convey their

nationalistic pride, similar to the Tunisians as suggested earlier.

Statistical Analysis

In this section, I have tried to show that the host used features that are more

colloquial with the Egyptians, accommodating to their colloquial style. I have also

shown that for the variable (ḍ), the on-the-street Tunisians used the colloquial variant

[ð̣] more than their in-studio counterparts, and that for the variable (ð̣), the Egyptians

used the colloquial variant [ẓ] more than the other speakers. Usually, the MANOVA test

or regression analysis is used to test the null hypothesis in a sociophonetic study.

However, the small number of speakers and tokens gathered from them, as well as the

use of categorical data prevented such an analysis.

Furthermore, the fact that certain speakers used both variants made it

impossible to code the speakers for statistical analysis, which requires putting a speaker

in one group and not the other. For example, the Palestinian ‘Abdul Bari Al-‘Atwan

could have been coded as belonging to the [ḍ] group because he showed no variation
 63

on this variable. Similarly, the on-the-street Tunisians could have been coded as

belonging to the [ð̣] group because they did not use the other variant. However, Rachid

Al-Ghannouchi, an in-studio Tunisian, switched between the two and could therefore

not be coded as belonging to either. Individual differences like these prohibited

statistical analysis.

Lexical and Syntactic Variables

Now that we have seen how variation occurs on the phonological level of the

language, we can move to the lexical and syntactic level to explore the interactions of

variation and style shifting there. This investigation will provide a more complex view

of style shifting by illustrating the intricacies of how various speakers manipulate their

language in different ways and in different settings. First, I will examine whether the

speakers chose to employ fuṣḥā or ʕāmijjə markers of negation, then I will investigate

whether they chose fuṣḥā or ʕāmijjə relative pronouns.

Negation

As discussed earlier, in fuṣḥā there is a distinction between nominal negation,

where ‫ ﻟﻴﺲﺱس‬lejsə is used, and verbal negation where various particles are used depending

on the tense of the verb. ‫ ﻟﻢﻡم‬lɛm and ‫ ﻤﻣﺎ‬mā are used for the past tense, ‫ ﻻ‬lā is used for the
 64

present, and ‫ ﻟﻦﻥن‬lɛn for the future, whereas in the dialects ‫ ﻤﻣﺶﺵش‬mɪš/muš, or the circumfix ‫ﻤﻣﺎ‬

‫ ﺶﺵش‬ma-š are used.

Tokens of each of these particles of negation were collected for all the speakers

on the program, and the results are represented in Table 5.

Percentages were calculated and are illustrated in Figure 6.

In Figure 6, again we can see that the Egyptians are leading the way by using

the most ʕāmijjə particles of negation compared to the other groups. These results are in

line with their tendency to use more ʕāmijjə markers than the other speakers and use an

overall more colloquial style. The next group after the Egyptians is the on-the-street

Tunisians, which is interesting for a number of reasons. First, this adds another

dimension to how the on-the-street Tunisians were using more ʕāmijjə characteristics

than the in-studio Tunisians. Just as they used the ʕāmijjə [ð̣] variant of [ḍ] more than

the in-studio Tunisians, they also use more ʕāmijjə particles of negation, thereby using a

more colloquial style.

The fact that Siham Bensedrine, the Tunisian woman on the phone, is next in

line for using more ʕāmijjə characteristics also illustrates another way she is using more

ʕāmijjə characteristics in general, which can be related to the relaxed domain of the

home.
 65

The fact that the various dialects and the in-studio Tunisians, both groups

entirely composed of males, use predominantly fuṣḥā particles of negation also

illustrates how the in-studio guests use a more standard style due to the formality of the

in-studio domain. And of course, the host maintains his role as moderator and Al-

Jazeera reporter by not using any ʕāmijjə particles of negation.

Table 5. Counts of negation particles produced by host and guests

fuṣḥā % ʕāmijjə % Total

Egyptians 8 15% 44 85% 52

Tunisian woman on-the-phone 11 69% 5 31% 16

On-the-street Tunisians 35 62% 21 38% 56

In-studio Tunisians 53 98% 1 2% 54

Various dialects 39 95% 2 5% 41

Host 101 100% 0 0% 101

Total 247 77% 73 23% 320

 66

Negation
100%
90%
80%
70%
60% fuṣha
50%
40% ʕamijjə
30%
20%
10%
0%

Figure 6. Particles of negation variation.

Relative Pronouns

As mentioned earlier, in fuṣḥā there are three relative pronouns depending on

gender and number of the noun being referred to: allɛðī for masculine singular, allɛtī for

feminine singular, and allɛðīnə for plural, whereas in the dialects ɪllī is used.

Tokens for these particles were collected for each speaker and the results are

represented in Table 6.
 67

Percentages were calculated and are illustrated in Figure 9.

Figure 7 is particularly interesting because it breaks the trend we have seen in the other

variables. Instead of the Egyptians leading the way, Siham Bensedrine (the Tunisian

woman on-the-phone) uses the most ʕāmijjə relative pronouns, followed by the

Egyptians. Interestingly, the on-the-street Tunisians are next in line for using the most

colloquial relative pronouns, which falls in line with the fact that they were using more

particles of negation from the colloquial.

The fact that the in-studio Tunisians and the speakers from various dialects also

used these relative pronouns adds another layer to the complexity of how the in-studio

guests sometimes flavor their predominantly fuṣḥā speech with some markers of ʕāmijjə.

Table 6. Counts of relative pronouns produced by host and guests

fuṣḥā % ʕāmijjə % Total

Egyptians 6 24% 19 76% 25

Tunisian woman on-the-phone 4 20% 16 80% 20

In-studio Tunisians 34 68% 16 32% 50

Host 71 92% 6 8% 77

Various dialects 17 85% 3 15% 20

On-the-street Tunisians 9 30% 21 70% 30

Total 141 64% 81 36% 222

 68

Relative Pronoun "that"

100%
90%
80%
70% fuṣha (allɛðī,
60% allɛtī, allɛðīnə)
50% ʕamijjə (ɪlli)
40%
30%
20%
10%
0%

Figure 7. Relative pronoun variation.

For example, ‘Abdul Bari Al-‘Atwan, the Palestinian journalist, showed no variation on

the phonological variables (ḍ) and (ð̣), maintaining the fuṣḥā pronunciations the whole

time. However, he used the ʕāmijjə relative pronoun ɪllī three times. The exploration of

variables on all levels of the language allows us to see that although he did not show

any variation on the phonological level, he did switch between fuṣḥā and ʕāmijjə

relative pronouns, adding another dimension to style shifting in Arabic.

 69

Perhaps the most interesting part of this graph is that unlike the particles of

negation, the host, Ghassan Bin Jiddu, uses some ʕāmijjə relative pronouns. Here, the

setting is important because he only used these ʕāmijjə relative pronouns when he was

on the street interviewing Tunisians. He never used ɪllī in studio. We can interpret this

as both style shifting and speech accommodation because he is using a more colloquial

style on the street, which again could be related to the cultural aspect of Arab

hospitality and making his guests feel comfortable to speak naturally and openly. He is

also accommodating to the Tunisian dialect. In the entire data set, he used ɪllī six times

out of seventy-one, and interestingly, half of these occur in his introduction to the

segment of the program out on the street with the Tunisians where he is setting up the

discussion. He reminisces about his college days in the seventies and eighties in Tunisia

when Tunisian universities had some of the most active student movements, be they

leftist, Islamic, or any type of movement where the students were involved and up to

date on current events. In this section of speech, he also uses the ʕāmijjə bi- prefix for

the present tense. He did not use this prefix anywhere else in the data set. Here he is

talking to this group of men as a fellow Tunisian and the purpose of this reminiscing is

to portray himself as one of the guys. By using ɪllī and the bi- present tense prefix, he is
 70

signaling the Tunisian aspect of his identity in order to establish solidarity with the men

he is about to interview.

Advantages

Now that I have presented the results for the variables in this study, I will

discuss the advantage of the approach taken here, as well as the disadvantages. The

broadcast media context provides some advantages that are not present in a typical

sociolinguistic interview. Usually in a sociolinguistic interview, the researcher will

inform the participant that they are being interviewed for the sake of linguistic

research. The presence of the recorder can make the participant more aware that the

researcher is focusing on their speech and how they are saying things. Furthermore, the

fact that they are being interviewed in the first place heightens this awareness, because

as an average citizen, they probably are not recorded or interviewed very often.

On this program, the guests have not been invited to the program for the sake of

listening to how they speak. They are on the program to voice their opinion about the

Arab Spring and the revolutions in their country. They are paying less attention to their

speech than a participant in a sociolinguistic interview; therefore, they may focus more

on what to say instead of how to articulate it, which may be more similar to their

unobserved speech. Furthermore, they have a vested interest in the topic because it
 71

immediately concerns them and their futures, therefore they will probably be authentic

and passionate about the subject.

Compared to a sociolinguistic interview where the audience is the academic

researcher, the guests on this program are tailoring their speech to both the host and

the pan-Arab audience watching Al-Jazeera. This context creates an interesting dynamic

to explore the interaction of whom the speaker is addressing, conversation setting and

topic, and the desire to be mutually understood.

Limitations

However, there are also some limitations of examining style shifting between

fuṣḥā and ʕāmijjə in the media context. First, there is not as much personal or

biographical information available compared to a sociolinguistic interview where the

participants respond to a questionnaire about their social and educational background.

This lack of biographical information prevents an investigation of the “social”

dimension of linguistic variation mentioned earlier. For example, without this

biographical information we cannot compare the results according to education or

socioeconomic status. Interviews and questionnaires are extremely useful in

sociolinguistic studies because they permit the researcher to gather personal

information about the participant.

 72

On the other hand, identity is highly complex, and even if one were to identify

oneself as “Egyptian”, this does not automatically entail that they will categorically

employ every feature of the Egyptian dialect in all settings and circumstances. For

example, one of the prominent features of Egyptian (and most urban dialects) is the

pronunciation of /q/ as [ʔ]. Although the Egyptians in my data do not use this variant

consistently, it does not make them any more or less Egyptian because nationality

cannot be taken as the sole determiner of linguistic variation.

Another issue with the data set is the fact that the guests of the program were

invited to participate because of their supposed knowledge or insight into the subject,

which means we are dealing with educated professionals who may not represent their

speech community as a whole. In the case of the Tunisians in the data set, the fact that

most of the in-studio Tunisians are journalists or politicians is partially mitigated by the

third episode where average Tunisians are interviewed on the streets. Most

sociolinguistic studies on the Arabic dialects usually examine educated speakers from

the higher classes, with a notable exception of Anne Royal’s (1985) investigation of

differential patterns of pharyngealization among males and females in two lower class

Cairene neighborhoods. Yet, more fieldwork and research is needed that looks at the

speech of lower class Arabs and rural areas.

 73

A final issue is that the television context is somewhat artificial due to the

presence of the cameras, which may exert influence over the speakers and how aware

they are of their own speech. This issue is somewhat related to the observer's paradox

because the presence of an observer (the camera, and by extension, the viewing

audience) influences how they may speak. Since they are on television, they also may

try to depict a persona that may be different from their own personality in reality.

These issues may limit the generalizability of some of the findings, but I think

the results still illuminate some of the complexities in the techniques that Arabic

speakers employ in switching between different speaking styles in different settings.

Conclusion

This sociolinguistic study has demonstrated the complexities of style shifting in

Arabic in the media setting through analysis of phonological, lexical, and syntactic

variables. Quantitative analysis of the variable ‫( ﺾﺽض‬ḍ) > ‫[ ﻈﻇﻆﻅظ‬ð̣] showed that the in-studio

Tunisians used the fuṣḥā variant [ḍ] more than the on-the-street Tunisians, who only

used the ʕāmijjə variant [ð̣]. The fact that some speakers like the host and the in-studio

Tunisians switched between both variants in the formal domain violated the functional

differentiation proposed in the diglossic framework.

 74

For the variable (ð̣), the Egyptians used their dialectal [ẓ] variant more than the

other guests. This was in line with their tendency to use more colloquial markers in

general. The Egyptians also used more ʕāmijjə particles of negation and relative

particles than the Tunisians or other dialect groups, although the Tunisian woman on

the phone used more ʕāmijjə relative pronouns. The in-studio Tunisians maintained

more fuṣḥā features, perhaps due to the formal studio domain, whereas the on-the-street

Tunisians maintained more ʕāmijjə features due to the casual and natural domain of the

street. The Egyptians maintained ʕāmijjə features even though they were in the formal

setting of the studio because their dialect is widely understood so they do not have to

level their dialect as much as the Tunisians.

A Closer Look at Diglossia and Code Switching

The results presented above illustrate the interaction of various styles of Arabic

in an array of settings for a diverse group of speakers. It shows the weakness of a binary

view of language (diglossia/code-switching) and highlights why a fluid, dynamic,

variationist framework is required. The data for the phonological variables showed that

speakers do not strictly use one variant over the other when speaking in a particular

style, and the data for the relative pronouns and particles of negations illustrated how

speakers switch between different styles on other levels of the language.

 75

Now that we have seen how the variables in this study were actually used, let us

return to the questions the results raise theoretically and methodologically about the

shared and intermediate forms. The lexical and syntactic variables are easy to analyze

because each item belongs to one variety and not the other. For example, ɪllī and mɪš do

not exist in fuṣḥā. Regarding the question of whether allɛðī and lɛn exist in ʕāmijjə, I will

assume these lexical items are shared. The phonological level, however, is more

complicated because here we have two points of differentiation between the two

varieties: individual phonemes and syllable structure. For example, a Tunisian can

choose whether to use the [ḍ] or [ð̣] variant, and they can also choose whether to use

CV (fuṣḥā) or CC (ʕāmijjə) syllable structure. Therefore, when they want to say a word

like ‫ ﺒﺑﻌﺾﺽض‬baʕḍ ‘some’, they can choose one phoneme, [ḍ] or [ð̣], and they can choose one

syllable structure baʕḍ (CV) vs bʕḍ (CC) over the other. They would then have four

possibilities

(1) baʕḍ (2) baʕð̣ (3) bʕḍ (4) bʕð̣

(1) is strictly fuṣḥā, maintaining standard phonemes and syllable structure. (2) could be

described as “colloquialized-fuṣḥā” under Boussafara-Omar’s model on the phonemic

level. (3) could also be “colloquialized-fuṣḥā” because of the change in syllable

structure, and (4) would be strictly ʕāmijjə. Perhaps it would be best to think of (2) and
 76

(3) as shared or neutral forms because it is difficult to classify them as strictly belonging

to one variety and not the other with such a minor change in phonology.

This issue is further complicated when other morphemes are added such as the

definite article. In fuṣḥā the definite article ‫ ﺍاﻟـ‬is usually pronounced as al-, whereas in

many of the dialects it can be pronounced as ɪl- or ɛl-. Therefore, a Tunisian would face

four possibilities if they want to say ‫ ﺍاﻻﻨﻧﺘﻔﺎﻀﺿﺔ‬al-ɪntifaḍa ‘uprising’

(5) al-ɪntifaḍa (6) al-ɪntifað̣a (7) ɪl-ɪntifaḍa (8) ɪl-ɪntifað̣a

First, we must question if we are to deal with the definite article and the following word

as a whole or analyze each morpheme individually. If we are to deal with them as a

whole, (5) is strictly fuṣḥā because the fuṣḥā definite article and [ḍ] phoneme are

maintained. (6) would be colloquialized-fuṣḥā since the [ð̣] variant is used even though

the fuṣḥā definite article is maintained. However, if we are to deal with them separately,

in (6) the definite article would be classified as fuṣḥā whereas the noun itself would be

classified as colloquialized-fuṣḥā on the phonological level. Again, it is questionable if

these would really be interpreted as intermediate forms due to the change of one

phoneme. Therefore, I will consider them shared or neutral because they cannot clearly

argue for one code over the other. I will also classify each morpheme separately for the

sake of illustration.
 77

Let us look at an excerpt of speech from the actual data set. First, I will present

an excerpt from the speech of Khalid Yusef, the Egyptian director. Items that are shared

between fuṣḥā and ʕāmijjə are kept in regular type, items belonging strictly to fuṣḥā are

bolded and items belonging strictly to ʕāmijjə are italicized. The intermediate forms are

underlined.

hijjə al-quwə ad-dāfɪʕ bɪlā ādnā šak hijə aš-šəbāb, jaʕnī aš-šəbāb ɪllī
xarəʤ[g]ū jawm xamsə wa ʕšrīn da tārīxij-an hum ṭalīʕ hāð[z]ɪhɪ aθ[s]-
θ[s]awrə, fa-hijjə dī al-quwə ad-dāfɪʕ jaʕnī al-məḥān allɛtī kānə[ ] ja-ʕīš-hā
haʔ[ ]ulā aš-šəbāb ʕlā a[ɪ]l-mustawā as-sijāsī wa ʕalā a[ɪ]l-mustawā al-
ɪʤ[g]timaʕī hijjə allɛtī dəfʕt-hā l-lxarūʤ[g] jə-ṭāləbū bi-kərāmə li-hāð[z]ā al-
wəṭən, jaʕnī fī-hum nās kəmān ʕlā fɪkrə kānū ʕnd-u-hum wəð̣[ẓ]āʔ[ī]f wa ja-
q[ʔ]darū ja-ʕīšū kwejsīn wa lākɪn hum šaʕrū bi-ɪnə a[ɪ]l-wəṭən muḥān, fī fuqərāʔ
wa ɪllī bɪ-jə-zhafū ʕalā baṭun-hum kej jā-kulū fī ə fī al-qahər ɪllī mawʤ[g]ūd fī š-
šuwarɪʕ mā-hada-š jə-q[ʔ]dar jɪ-nṭəq[ʔ] mā-ḥada-š jə- q[ʔ]dar ju-ʕārɪḍ kul da
hum šaʕrū bi-hi ḥəttā law kānə[ ] lɛm ju-mārɪs ʕlā wāḥɪd mɪn-hum kānə[ ] ḥəttā
al-wāḥəd da ḥāsɪs da.

In this excerpt, we can observe the mix of varieties Khalid Yusef uses at his disposal in

the studio setting. We can see that he only uses one morpheme that belongs strictly to

fuṣḥā, the connector fa-, which here we can interpret as ‘then’ or ‘also’. In this excerpt,

he uses several morphemes from ʕāmijjə like the demonstrative da instead of fuṣḥā hāðā

for ‘this’. Interestingly, he uses both da and hāðā, except when he uses hāðā he uses his

dialectal phonology so that he says hāzā. This is an example of what Labov called
 78

‘inherent variability’ because he is using both fuṣḥā and ʕāmijjə demonstratives and

phonological variants in the same stretch of speech.

Perhaps most importantly, this sample speech illustrates why ignoring the

shared forms ignores so much data. After calculating the total number of each variety

that were used in the example above, I found that 87% of these morphemes were

shared. Only one could be identified as strictly fuṣḥā, eighteen as ʕāmijjə, and two as

intermediate forms. Both the intermediate forms occurred when he used the Egyptian

ʕāmijjə system of negation mā-š with a morpheme that belonged to both varieties. The

fact that Khalid Yusef used so many features of ʕāmijjə are in line with the results

presented for the variables above, where it was shown that the Egyptians used more

ʕāmijjə markers in general, thereby using a more colloquial style.

Compare Khalid Yusef’s speech to the following sample from the speech of

Rachid Al-Ghannouchi, a politician in the En-Nahda party in Tunisia who spoke to the

host in studio, on a panel with the Palestinian journalist ‘Abdul Bari Al-‘Atwan.

ʔanā ʔa-daʕū li-ʔakθar mɪn ðalɪk ʔa-daʕū ɪlā təfkīk mənð̣ūmət a[ɪ]l-ɪstɪbdād ʔa-
dʕū ɪlā ʔan, lā ʔa-ʕtəbr ʔal allɛðīnə šārək-ū fī nið̣ām a[ɪ]l-ɪstɪbdād fɪ-ṣəf al-ʔawəl
xāṣə-tan wa aṣ-ṣəf aθ-θānī hum qādɪrūn ʕlā bināʔ al-muʤtɛmʕ ad-dīmuqrāṭī
muʤtɛmaʕ al-ʕdəl wa al-ḥurijjə allɛðī rəmət ɪlej-h-i hāðihi al-ɪntifaḍ[ð̣]a wa
liðālɪk lā buddə mɪn al-ātiʤā-h-i ɪlā mənð̣ūmat al-ɪstiɪbdād qawānīn dasātīr wa
riʤāl wa muʔwəsəsāt l-tafkīk-hā mɪn ʔaʤal-i banāʔ dawilə li-tūnɪs wa lejsə
dawilə l-ʕāʔɪlə wa lā dawilət l-bolīs, hāðā xəṭər ḥəqīqī wa liðālɪkə ʔanā ʔa-daʕū
ʔahəl al-ɪntifāḍ[ð̣]ə ʔa-daʕū šaʕb tūnɪs ɪlā təmām-i al-jəqəð̣ə wa tamām al-
 79

ḥuḍ[ð̣]ūr ḥattā jə-hamī θawrətə-hu[ ] wa ɪntifāḍ[ð̣]atə-hu wa ḥattā ju-ḥāfɪð̣ ʕlā

dimāʔ aš-šuhadāʔ wa ʔamal hāðā aš-šaʕb.

Here we can see that Rachid Al-Ghannouchi uses far fewer colloquial forms than Khalid

Yusef. He also used three markers of fuṣḥā, the genitive case marking –i. These three

morphemes only constituted two percent of the data set, because the other 98% was

shared items. From this example, we can also see why the (ḍ) variable was chosen,

because it is the most salient phonological variation in his speech. Overall, we can say

that Rachid Al-Ghannouchi used many features that are shared between the varieties

and formal fuṣḥā case endings in a few places, but he also maintained his ʕāmijjə [ð̣]

variant of (ḍ).

Compare the speech of Rachid Al-Ghannouchi to that of Samir Ben ‘Ali, a

Tunisian man who was interviewed on the street, which is presented below:

ʕnd-ī šəqīq huwwə ɪllā jɛ-ʤīb l-ī a[ɪ]l-faṭūr huwwə jɛʤīb l-ī a[ɪ]l-ʔakal jaʕnī
kānɛt[ ] ð̣urūf-nā qāsijə kānə[ ] jaʕnī ju-ʤīnī kānə[ ] baʕd məšəqə jaʕnī məšəqə
kabīrə, kānɛt[ ] aš-šurṭə tə-bḥɛθ ʕan-ī fī kul mɛkān, ʕnd-ī xaṭībə
kunta[ ] xāṭɛb-hā jaʕnī mā rɪmdū bi-hālat-hā jaʕnī mā rɪmdū θəmānī sɛnuwāt
mɪn ʔaʤɪl ʔan-hā xaṭībt-ī tə-bḥəθ fī wa ɪtəsɛʤnɛt wa xarɪʤt[ ] mɪn as-sɛʤɪn
mašī[ɪ]t ʕamɪl-ū lɪ-hā bɪdūn bɪdūn jaʕnī qərār qaḍ[ð̣]aʔ[ī] ʕmɪl-ū lɛ-hā a[ɪ]l-
murāqibə a[ɪ]l-ɪdārijə θāmanī sɛnuwāt ɪllī mɛlɛt jaʕanī mɛlɛt mɛlɛt jaʕanī təfārəq-
at ʔanā ʔijāhā qult[ ] lɛ-hā ɪmši ʔɪnti šūf-ī hal wa mā-ʕāda-š nə-nɛʤəm nə-bqə fɪl-
wəḍ[ð̣]ʕ hāðā ɪllī bāš jə-kar ʕlā al-kul…ʔɪlā al-ān, təwə bāš jə-təfāʤʔū ʔɪðā
kānə[ ] jə-šūfū-nī jaʕanī fi at-tɪlfāz bāš jə-təfāʤʔū, ʔaxuwī jə-qūl kɪðā šakūn, mā-
jə-ʕrəfūnī-š. hāðihi ʔawəl marrə jaʕanī nə-šūf.

Compared to Khalid Yusef and Rachid Al-Ghannouchi, Samir Ben ‘Ali used a large
 80

number of ʕāmijjə markers, 23 in all, which was 14% of his speech in this excerpt. Most

of these ʕāmijjə morphemes are lexical items that belong strictly to Tunisian Arabic like

‫ ﺒﺑﺎﺶﺵش‬bāš ‘because’ instead of the fuṣḥā equivalent ‫ ﻷﻦﻥنﱠ‬liʔɛnnə. Again, we can see that the

majority of his speech was shared morphemes, 85% in all. There are two intermediate

forms in this excerpt where he uses the ʕāmijjə mā-š construction of negation around a

lexical item that is shared in both varieties. Interestingly, he uses the same expression ‫ﻤﻣﻦﻥن‬

‫ ﺃأﺠﺟﻝل‬mɪn ʔaʤɪl ‘for the sake of’ or ‘in order to’ that Rachid Al-Ghannouchi used, but

without the fuṣḥā genitive case marker –i like Rachid. This could be due to lack of

education in fuṣḥā, or it could be due to the more colloquial style he is using in the

informal domain of the street.

Looking at the data visually in this way illustrates why it is difficult to classify

certain items as belonging to one variety and not the other, and how pervasive these

items are in the language. It also allows us to see the general trends discussed above:

how the Egyptians used several features of their dialect, and the on-the-street Tunisians

used more ʕāmijjə markers than their in-studio counterparts. I believe it is best to look

at this data from the perspective of switching between various styles.

 CHAPTER 6

CONCLUSION

This study has investigated linguistic variation on the phonological, lexical, and

syntactic levels of the language, demonstrating how Arabic speakers have various

resources at their disposal to shift between styles. In the process, I have hoped to show

how the diglossic framework is not useful to explain linguistic variation in Arabic

because actual spoken Arabic contains various combinations of both varieties.

Future Research

This study is by no means a comprehensive account of the linguistic variation in

these three episodes or Egyptian and Tunisian Arabic. The Egyptians in particular

exhibited much variation on other phonological variables like ‫( ﻖﻕق‬q) and [ʔ], ‫( ﺬﺫذ‬ð) and ‫ﺯز‬

[z], and ‫( ﺚﺙث‬θ) and ‫[ ﺖﺕت‬t] or ‫[ ﺲﺱس‬s], however these variables were outside the realm of this

study. The goal of this study was not to characterize every instance of phonological

variation in Egyptian because there is an abundance of materials in the literature on EA.

 82

The field of Arabic linguistics needs more investigations into linguistic variation

of various dialects throughout the Arab world, particularly Tunisian Arabic since so few

resources were found for this dialect. As stated earlier, there is much research to be

done for variants of (ḍ) in all the dialects.

There is also a need for more studies on less educated speakers. Many

sociolinguistic studies in the literature have been carried out in the university setting,

and more fieldwork has to be done on more speech communities in rural and poorer

areas. Unfortunately, this study also fell into this tendency to focus on educated

speakers, which was mainly due to the availability of the transcripts and the media

setting.

I have tried to show how analyzing the linguistic complexities of variation in

Arabic are best handled by the style-shifting framework. The various theories that have

been proposed to fix the inadequacies of the diglossic framework were insufficient to

explain or predict actual speech data. I hope that this study has provided a more

complete view of linguistic variation that will be of use to linguists and students of

Arabic alike.
 ‫‪APPENDIX‬‬

‫‪TRANSCRIPTION SYSTEM‬‬

‫‪Consonants‬‬

‫ﺃأ‬ ‫‪ʔ‬‬ ‫ﺾﺽض‬ ‫‪ḍ‬‬

‫ﺐﺏب‬ ‫‪b‬‬ ‫ﻄﻃﻂﻁط‬ ‫‪ṭ‬‬
‫ﺖﺕت‬ ‫‪t‬‬ ‫ﻈﻇﻆﻅظ‬ ‫̣‪ð‬‬
‫ﺚﺙث‬ ‫‪θ‬‬ ‫ﻉع‬ ‫‪ʕ‬‬
‫ﺝج‬ ‫‪ʤ‬‬ ‫ﻍغ‬ ‫‪ɤ‬‬
‫ﺡح‬ ‫‪ḥ‬‬ ‫ﻒﻑف‬ ‫‪f‬‬
‫ﺥخ‬ ‫‪x‬‬ ‫ﻖﻕق‬ ‫‪q‬‬
‫ﺪﺩد‬ ‫‪d‬‬ ‫ﻚﻙك‬ ‫‪k‬‬
‫ﺬﺫذ‬ ‫‪ð‬‬ ‫ﻝل‬ ‫‪l‬‬
‫ﺭر‬ ‫‪r‬‬ ‫ﻢﻡم‬ ‫‪m‬‬
‫ﺯز‬ ‫‪z‬‬ ‫ﻦﻥن‬ ‫‪n‬‬
‫ﺲﺱس‬ ‫‪s‬‬ ‫ﻩه‬ ‫‪H‬‬
‫ﺶﺵش‬ ‫‪š‬‬ ‫ﻮﻭو‬ ‫‪W‬‬
‫ﺺﺹص‬ ‫‪ṣ‬‬ ‫ﻱي‬ ‫‪J‬‬
 84

Long Vowels

‫ﺍا‬ ā
‫ﻱي‬ ī
‫ﻮﻭو‬ ū

Short Vowels

i ɪ ɛ æ a ə u
 REFERENCES

Abdel-Jawwad, H. (1981). Lexical and phonological variation in spoken Arabic in

Amman. PhD thesis, University of Pennsylvania.

Abdel-Jawwad, H. (1987). Cross-dialectal variation in Arabic: Competing prestigious

forms. Language and Society, 10, 359-367.

Abudalbuh, M. (2010). Effects of gender on the production of emphasis in

Jordanian Arabic: A sociophonetic study. (Unpublished master’s thesis).
University of Kansas.

Abu-Haidar, F. (1987). Are Iraqi women more prestige conscious than men?
Sex differentation in Baghdadi Arabic. Language in Society, 18, 471-481.

Al-Jazeera. (2000, January). ‫ ﺍاﻟﻘﻨﻧﺎﺓة ﺒﺑﺮﻨﻧﺎﻤﻣﺞ‬- ‫ﻤﻣﻔﺘﻮﻭوﺡح ﺤﺣﻮﻭوﺍاﺭر‬. In ‫ ﺍاﻟﺠﺟﺰﻳﺮﺓة‬- ‫ﺍاﻟﻔﻀﺿﺎﺌﺋﻴﺔ‬.

Retrieved December 22, 2012, from
http://www.aljazeera.net/programs/27d796cb-abce-444f-a271-24c83bfc051c

Al-Khatany, A. (1997). The problem of diglossia in the Arab world. Al-‘Arabiyya, 30, 1-
30.

Al-Masri, M. (1998). The acoustic and perceptual correlates of emphasis in

urban Jordanian Arabic. (Unpublished doctoral dissertation). Yarmouk
University, Jordan.

Al-Wer, E. (2007). The formation of the dialect of Amman: From chaos to order. C.
Miller, E. Al-Wer, D. Caubet and J. Watson (Eds.), Arabic in the city: Issues in
dialect contact and language variation (pp. 55-76). Abingdon: Routledge Curzon.

Al-Wer, E. (2002). Education as speaker variable. In Rouchdy A. (Ed.) Language

contact and language conflict in Arabic (pp. 41-53). Curzon: London.
 86

Amor, T. B. (1990). A beginner’s course in Tunisian Arabic. Peace Corps, Tunis

(Tunisia).

Badawi, M. M. (1973). Mustawayaat al-‘arabiyya al-mu’aasira fi Misr. Cairo: Daar

al-Ma‘arif.

Bahloul, M. (2007). Linguistic diversity: The qaaf across Arabic dialects. In

Benmamoun (Ed). Perspectives on Arabic linguistics (pp. 247-266). XIX Urbana:
Benjamins.

Bakir, M. (1986). Sex differences in the approximation to standard Arabic: A case

study. Anthropological Linguistics, 28,1, 3-9.

Bassiouney, R. (2006). Functions of code switching in Egypt: Evidence from monologues.

Leiden: Brill.

Bassiouney, R. (2009). Arabic sociolinguistics: Topics in diglossia, gender, identity, and

politics: Georgetown University Press.

Baoueb, L. B. (2009). Social factors for code-switching in Tunisian business

companies: A case study. Multilingua, 28, 425-458.

Belazi, H. M. (1991). Multilingualism in Tunisia and code-switching among educated

Tunisian bilinguals. Doctoral dissertation, Cornell University, Ithaca, N.Y.

Bell, A. (1984). Language style as audience design. Language in society, 13, 145-204.

Bin Jiddu, G. (2011, April 29). In Facebook: ‫ﺠﺟﺪﺩدﻮﻭو ﺒﺑﻦﻥن ﻏﺴﺳﺎﻦﻥن‬. Retrieved February 6, 2012,
from https://www.facebook.com/Ghassan.ben.jiddo

Blanc, H. (1960). Stylistic [style] variations in spoken Arabic: A sample of

interdialectal educated conversation. In C. A. Ferguson (Ed.) Contributions to
Arabic linguistics (pp. 81-156). Cambridge, MA: Harvard University Press.

Blanc, H. (1964). Communal dialects in Baghdad. Cambridge Mass.: Harvard

University Press.
 87

Boxberger, L. (1981). Acoustic characteristics of Arabic pharyngeal and

pharyngealized consonants. Kansas Working Papers in Linguistics, 6,
127-152.

Boussofara-Omar, N. (1999). Arabic diglossic switching in Tunisia: An

application of Myers-Scotten’s MLF Model. (Unpublished doctoral
dissertation). University of Texas: Austin.

Bouzemmi, A. (2005). Linguistic situation in Tunisia: French and Arabic code-

switching. Interlinguistica, 16, 1, 217-223.

Daher, J. (1998). Gender in linguistic variation: The variable (q) in Damascus

Arabic. In Elabbas Benmamoun, Mushira Eid and Niloofar Haeri (Eds.).
Perspectives on Arabic Linguistics (pp. 183-208) XI Amsterdam: Benjamins.

Daher, J. (1999). (θ) and (ð) as ternary and binary variables in Damascene Arabic.
Perspectives on Arabic Linguistics XII, 163-202.

Daoud, M. (2011). The sociolinguistic situation in Tunisia: language rivalry or

accommodation?. International Journal of the Sociology of Language, 211, 9-33.

Eddington, D. and M. Taylor. (2009). T-glottalization in American English. American

speech. 84, 3, 298-314.

Eddington, D. and C. Channer. (2010). American English has goʔ a loʔ of glottal stops:
Social diffusion and linguistic motivation. American speech, 85, 3, 338-351.

Eid, M. (1982). The non-randomness of diglossic variation in Arabic. Glossa, 16, 54-84.

Eid, M. (1988). Principles for code-Switching between standard and Egyptian Arabic.
Al-‘Arabiyya, 21, 51-80.

El-Hassan, S. (1977). Educated spoken Arabic in Egypt and the Levant: A critical review
of diglossia and related concepts. Archivum linguisticum, 8, 2, 112-132.

Ferguson, C.A. (1959). The Arabic koine. Language, 25, 616-630

 88

Ferguson, C. A. (1959). Diglossia. Word, 15, 325-40.

Ferguson, C.A. (1991). Epilogue: diglossia revisited. Southwest Journal Linguist,

10, 1, 214-234.

Gibson, M. (2002) Leveling in Tunisian Arabic: Towards a new standard. In Aleya

Rouchdy (Ed.), Language contact and language conflict in Arabic: Variations on
a sociolinguistic theme (pp. 24-40.). New York: Routledge Curzon.

Giles, H. and P. Smith. (1979). Accommodation theory: Optimal levels of

convergence. In Giles, H. and R. N. St. Clair (Eds.), Language and Social
Psychology (pp. 45-65). Oxford: Blackwell.

Haeri N. (1987). Male/female differences in speech: An alternative interpretation.

Variation in language NWAV-XV at Stanford: Proceedings of the fifteenth
annual conference on new ways of analyzing wariation [1986] ed. by Keith M.
Denning et. al. (pp. 173-182). Stanford, CA: Department of Linguistics, Stanford
University.

Haeri, N. (1996). The sociolinguistic market of Cairo: Gender, class, and education.
London: Kegan Paul.

Hachimi, A. (2007). Becoming Casablancan: Fessis in Casablanca as a case study. In C.

Miller, E. Al-Wer, D. Caubet and J. Watson (Eds.), Arabic in the city: Issues in
dialect contact and language variation (pp. 97-122). Abingdon: Routledge Curzon.

Havelova, A. (2000) Sociolinguistic description of Nazareth. In M. Mifsud (Ed.)

Proceedings of the third international conference of AIDA (pp. 141-44), Hal Lija:
Association international de dialectologie arabe (AIDA).

Holes, C. (1995). Community, dialect and urbanization in the Arabic-speaking

Middle East. Bulletin of the school of oriental and African Studies, 58, 2, 270-287.

Holes, C. (2004). Modern Arabic: Structures, functions, and varieties. Washington

D.C.: Georgetown University Press.
 89

Ibrahim M. (1986). Standard and prestige language: a problem in Arabic

sociolinguistics. Anthropological linguist, 28, 115-26.

Jabeur, M. (1987). A sociolinguistic study in Tunisia: Rades. PhD. Dissertation,

University of Reading: Reading, England.

Khawly, M. (2012, June 12). Ahmad ismail al-sayyed: Art as a weapon in Egyptian
revolution. In Alakhbar English. Retrieved September 25, 2012, from
http://english.al-akhbar.com/node/8516

Labov, W. (1963). The social motivation of sound change. Word, 19, 272-309.

Labov, W. (1969). Contract, deletion, and inherent variability of the English copula.
Language, 45, 715-762.

Labov, W. (1972). Some principles of linguistic methodology. Language in Society, 1, 97-

120.

Lawson-Sako, S. and I. Sachdev. (2000). Code-switching in Tunisia: Attitudinal and

behavioral dimensions. Journal of pragmatics, 32, 1343-1361.

Lewis, M. P. (Ed.), (2009). Ethnologue: Languages of the world, Sixteenth edition.

Dallas, Tex.: SIL International. www.ethnologue.com.

Lynch, M. (2007). Arab arguments: Talk shows and the new Arab public sphere. In P.
Seib New media and the new Middle East (pp. 101-117). Palgrave: New York.

Mejdell. G. (2006). Mixed styles in spoken Arabic in Egypt: Somewhere between order and
chaos. Leiden: Brill.

Meiseles, G. (1980). Educated spoken Arabic and the Arabic language continuum.
Archivum linguisticum, 11, 2, 118-148.

Miller, C. (2004). Variation and change in Arabic urban vernaculars. In M.

Haak, R. De Jong, and K. Versteegh (Eds.), Approaches to Arabic dialects: A
 90

collection of articles presented to Manfred Woidich on the occasion of his sixtieth

birthday (pp. 177-206). Leiden: Brill.

Mitchelle, T. F. (1978). Educated spoken Arabic in Egypt and the Levant, with special
reference to participle and tense. Journal of linguistics, 14, 2, 227-258.

Mitchell, Terrence F. (1986). What is educated spoken Arabic? International Journal

of the Sociology of Language, 61, 7-32.

Muqbil, M. (2006). Phonetic and phonological aspects of Arabic emphatics and

gutterals. (Unpublished doctoral dissertation). University of Wisconsin,
Madison.

Myers-Scotton, C. (1993). Social motivations for codeswitching: Evidence from Africa.

Oxford: Clarendon.

Negm, N. (2006, December 28). In ‫ ﺍاﻟﺸﺷﻌﺒﺑﻴﺔ ﺍاﻟﺘﻬﮭﮫﻴﻴﺲﺱس ﺠﺟﺒﺑﻬﮭﮫﺔ‬A popular front of sarcasm.
Retrieved February 6, 2012, from http://tahyyes.blogspot.com/

Roberts, J. (2006). As old becomes new: Glottalization in Vermont. American speech,

81, 3, 227-249.

Royal, A. (1985). Male/female pharyngealization patterns in Cairo Arabic: A

sociolinguistic study of two neighborhoods. Austin, TX: Texas Linguist. Form 27.
University of Texas, Austin.

Sallam, A. (1980). Phonological variation in educated spoken Arabic. Bulletin of the

School of Oriental and African Studies, 43, 77-100.

Sayahi, L. (2011). Code-switching and language change in Tunisia. International

Journal of the Sociology of Language, 211, 113-133.

Soliman, A. (2008). The changing role of Arabic in religious discourse: A sociolinguistic

study of Egyptian Arabic. (Unpublished doctoral dissertation). Indiana
University of Pennsylvania.
 91

Talmoudi, F. (1980). The Arabic dialect of Susa (Tunisia). Gothenburg: Acta

Universitatis Gothovergensis.

Walters, K. (2003). Fergie’s prescience: The changing nature of diglossia in

Tunisia, International Journal of the Sociology of Language, 163, 77-109.

Watson, J. (2002). The phonology and morphology of Arabic. Oxford: Oxford

University Press