The Arabic Language

Chapter 13
Diglossia
13.1 The nature of diglossia
In written Arabic, the choice between the standard norm and the colloquial 
language at first sight appears to be relatively uncomplicated: in writing, Standard 
Arabic is always used. But even here, problems of selection may arise. One compli
-
cation derives from the fact that many people possess only a limited knowledge 
of the standard norm. For these people, Standard Arabic remains the target, but 
in writing it they make many mistakes. This results in the so-called Middle Arabic 
texts that have been discussed above (Chapter 9). A further complication may 
arise when, for ideological or literary reasons, writers decide to compose their 
literary writings in an approximate version of the colloquial language. Even these 
authors usually mix their colloquial language with elements from the standard 
language.
In spoken Arabic, the situation is far more complicated. Perhaps the best 
analogue to the situation in the Arabic-speaking countries would be that of a 
hypothetical modern France, where all newspapers and books are written in 
Latin, speeches in parliament are held in Latin, and in churches the only language 
used by the priests is Latin. On the other hand, people talking in a cafe use French, 
people at home or among friends use French. In school, the official language of 
the classroom is Latin, but during the breaks between classes students use French 
among themselves, and so do the teachers. We know, of course, that the situation 
in France is not like this; but things could have been different, had the standard 
norm not switched from Latin to vernacular French in the fifteenth and sixteenth 
centuries.
In Arabic-speaking countries, the actual situation is very much like the 
hypothetical situation sketched here for France. At first sight, there appear to 
be two varieties of the language, the Classical standard, usually called 
fuṣḥā
, and 
the colloquial language, usually called 
ʿāmmiyya
or (in North Africa) 
dārija
, and in 
Western publications, ‘dialect’ or ‘vernacular’. These two varieties divide among 
themselves the domains of speaking and writing: the standard language is used for 
written speech and for formal spoken speech, whereas the colloquial language is 
used for informal speech. The colloquial language is everybody’s mother tongue; 

---

242
The Arabic Language
people learn the standard only when they go to school. In 1930, William Marçais 
called such a linguistic situation ‘diglossia’ (French 
diglossie
), a term that he had 
borrowed from the literature on the linguistic situation in Greece and which 
gained general currency after the appearance in 1959 of an article by Charles 
Ferguson with the title ‘Diglossia’. Ferguson compared the linguistic situation in 
the Arabic-speaking countries to that in Greece, in German-speaking Switzerland, 
and in Haiti. In all four areas, there is a similar functional distribution between 
two varieties of the same language (
fuṣḥā
/
ʿāmmiyya
, 
katharévousa
/
dhimotikí
, 
Hochdeutsch
/
Schwyzertüütsch
, 
français
/
créole
). In Ferguson’s terminology, these 
two varieties are called ‘high variety’ (H) and ‘low variety’ (L).
The terms ‘high’ and ‘low’ reflect the standing of the two varieties in the 
linguistic community. The low variety is held in very low esteem, and the name 
by which speakers refer to it normally implies a humble position: 
ʿāmmiyya
liter
-
ally means ‘common’ or ‘vulgar’, while other names are 
sūqiyya
‘language of the 
market’, 
munḥarifa
‘deviant’, and so on. The high variety, on the other hand, is 
prestigious: it is the language of a cultural, and often religious, heritage. Speakers 
may even deny the existence of the low variety and claim that they only speak 
the high variety. In reality, the low variety is the mother tongue of all speakers, 
whereas the high variety is a second language that is almost never used in impro
-
vised speech.
The theoretical framework of Ferguson’s model for the linguistic situation in 
the Arabic-speaking countries has been refined by subsequent studies in three 
important respects. In the first place, Ferguson’s model restricted the notion of 
‘diglossia’ to situations where the low variety was genetically related to the high 
variety, of which it was a reduced version. In later publications this restriction 
was lifted, and the notion of ‘diglossia’ was expanded to include any functional 
distribution of linguistic varieties, whether these were languages or dialects or 
registers. The functional distribution in the Arabic-speaking countries is merely 
a special case of a general phenomenon of sociolinguistic variation in all speech 
communities.
In the second place, the existence of a functional distribution between varieties 
does not imply that all speakers have an equal command of these varieties. In 
extreme cases, most speakers know only one variety, a non-prestigious colloquial 
kind of language, whereas a small elite uses a stilted variety of a cultural language, 
mostly an imported one. In the Arab world, an example of such a situation is 
Algeria just before independence: the majority of people in the speech commu
-
nity knew only Arabic and at the most a smattering of French, but a small group 
of intellectuals had been raised and educated in French, and had lost the ability 
to speak Arabic (see Chapter 14, p. 263). Several linguists, among them Fishman 
(1967, 1972) and Gumperz (1962), therefore proposed to distinguish between a 
sociolinguistic and a psycholinguistic approach. In their terminology ‘diglossia’ 
is reserved for the sociolinguistic notion of a functional distribution of linguistic 

---

Diglossia 
243
varieties. For the psycholinguistic notion of the speakers’ command of these 
varieties, they adopted the term ‘bilingualism’. In the literature, the specification 
‘societal bilingualism’ is sometimes used to refer to Ferguson’s use of the term.
A third modification of Ferguson’s model concerns the distinction of two 
discrete varieties. In his framework, the two varieties are mutually exclusive, 
and the speaker has to choose one or the other by a process of code-switching. 
In reality, the speaker never opts for one variety or the other, but moves along 
a continuum of speech, of which the two varieties are only the extremes. In 
such a situation, code-switching does not imply selecting a discrete variety, but 
positioning one’s utterance along a scale of linguistic variation. A better term 
to describe this kind of speech behaviour would perhaps be ‘code-mixing’, since 
there is no actual switch from one variety to another. Extralinguistic factors 
determine the position on this scale. Obviously, the span of the continuum that 
individuals control depends on their linguistic proficiency, which in its turn is 
determined to a large degree by their education and upbringing.
The use of the term ‘diglossia’ (or its Arabic synonym 
izdiwājiyyat al-luġa
) in 
the literature on the linguistic situation in the Arabic-speaking world has created 
some confusion. In Ferguson’s model, ‘diglossia’ is used only for the relationship 
between 
fuṣḥā
and 
ʿāmmiyya
, whereas the functional division between French 
and Arabic in North Africa was termed by him ‘bilingualism’. In this chapter, the 
term ‘diglossia’ will be used in the sense of a linguistic situation in which several 
speech varieties divide among themselves the domains of verbal communication. 
The term ‘bilingualism’ will refer to the individual’s proficiency in more than 
one speech variety (note that in Chapter 14, on the linguistic situation in the 
Maghreb, the term ‘bilingualism’ will be used in a slightly different way). In a 
speech community that is both diglossic and bilingual, there is a well-defined 
distribution of domains between more than one variety and all speakers are able 
to vary their linguistic behaviour between these varieties. We shall first look at 
code-mixing and the factors determining language choice in discourse, then at 
language attitudes and associations in the speech community, and finally at the 
correlation with intrapersonal variables.