Why Are There So Many Arabic Varieties?—And Traces of Roman Heritage in Moroccan Arabic

Apr 27, 2015 by

One of the question that keeps coming up in my online “Languages of the World” course is why languages diversify. As Ferdinand de Saussure explains in his Course in General Linguistics, language diversification is produced by the constant process of language change: languages always change but “no records indicate that any language has ever changed in the same way throughout its territory” (p. 199). Some changes are triggered by language-internal factors, while others result from contact with other languages.

map-arabic-dialectThe case of Arabic is particularly instructive because of its quick spread over vast territory in the 7th-9th centuries. The reason that Arabic was able to grow from a language of small nomadic tribes in the Arabian Peninsula to a language of nearly half a billion people from Morocco to Uzbekistan is language shift: entire groups speaking other languages, crucially including adults, abandoned their indigenous languages and began speaking Arabic instead. This linguistic conversion left an indelible mark on the vernacular varieties of Arabic (see map on the left), making them different both from the Modern Standard Arabic (MSA) and from each other.

Berber languagesLet’s consider Moroccan Arabic, a distinctive variety with its own peculiarities in the lexicon, sound system, and grammar. Before the arrival of Arabic, Morocco was inhabited by speakers of Berber languages. Berber and Semitic (a family to which Arabic belongs) are both branches of a bigger Afro-Asiatic language family, making Berber languages distant cousins of Arabic. However, the historical relationship between Berber and Arabic is so distant (going back at least 10,000 years) that they could as well be unrelated: adult Berber speakers were picking up a language that was quite foreign to them. Their linguistic conversion began in the military, with the troops composed mostly of Berber soldiers led by Arab commanders. Eventually, Arabic spread beyond the army, yet many Berber speakers retained their indigenous languages, which are still spoken in interior Morocco and elsewhere in northwestern Africa.

Interestingly, traces of Berber speakers’ linguistic conversion can be found in modern Moroccan Arabic. As John McWhorter notes in What Language Is, “[the Berbers’] rendition of Arabic was more meat-and-potatoes than the Classical one” (pp. 41-42). What he refers to is the relative grammatical (particularly, morphological) simplicity of Moroccan Arabic compared to MSA. For example, in MSA suffixes on nouns mark them for case (nominative, accusative, and genitive) and (in)definiteness: ‘a house’ (subject) is bayt-un, ‘the house’ (subject) is bayt-u, ‘the house’ (object) is bayt-a, and so on. No such suffixes are found in Moroccan Arabic. Similarly, verbs in MSA have forms that encode person (1st, 2nd, or 3rd), number (singular, dual, or plural), and gender (masculine or feminine). For example, ‘you two wrote’ is katabtumā, ‘you many guys wrote’ is katabtum, ‘you many gals wrote’ is katabtunna, ‘they many guys wrote’ is katabū, and so on. In contrast, in Moroccan Arabic, “the male-female and ‘just-two’ issues are long gone” (McWhorter 2011: 41), and only the person is marked on the verb: ktəbtu is ‘you people wrote’ (two or more, male or female), while kətbu is ‘they wrote’. As McWhorter argues in the book, such simplifications emerge whenever a group of adult learners picks another language: as McWhorter puts it, such adult learners “shear” excessive grammatical “bric-a-brac”, particularly morphological distinctions that lead to complex paradigms (as with MSA verbs). McWhorter half-jokingly notes that Moroccan Arabic is “the way we think of as normal (because Modern English is as ‘unnatural’ as Moroccan Arabic!)” (p. 41): both Moroccan Arabic and English were at some point picked up—and simplified—by grown-up speakers of other languages. (In the case of English, it was speakers of Celtic, Old Norse, and Norman French.)


But simplification in the hands (or rather, mouths!) of adult learners is only one way in which traces of the linguistic conversion can be perceived in Moroccan Arabic. As noted in Heath (2015), Berber was not the only language (or family of related dialects) spoken by people who ultimately converted to Arabic. According to him, the linguistic conversion occurred “in pre-existing Roman cities that were occupied by the first Arab-led soldiers and settlers”, particularly, Tangier, Salé and Volubilis (see map on the left, reproduced from Heath 2015: 3). The heyday of the Roman power in this area was in the late 3rd century, and by the 7th century “all of Roman North Africa was nominally ruled by the Byzantines” (Heath 2015: 4), yet “the Roman heritage was paradoxically more robust in sparsely populated Morocco than in the intensively colonized Tunisia” (ibid), and a significant proportion of the population in the three above-mentioned cities were “[Late Latin]-speaking Christian Romans”.

Unfortunately, remnants of the Roman heritage are hard to find in present-day Morocco. Archeological study is hampered by the fact that Roman cities, particularly Tangier, are buried under the modern metropolises. It is known that the presence of Christian Church in Morocco has dwindled by the beginning of 8th century, when the first sustained Arab garrisons are established in Tangier, Salé, and Volubilis. Nor do we have any documents in vernacular Late Latin dating from the period of Arab arrival in Morocco. (Nor do we have documents in Late Latin from Europe dating from the same time, as in Europe “serious Romance documentation begins only in 842 with the Strasbourg Oaths”; Heath 2015: 5.) Linguistic traces of Late Latin in modern vernacular Moroccan Arabic are, therefore, our best bet in discovering late Roman heritage in North Africa.

Heath (2015: 2) alludes to a number of phonological peculiarities of Moroccan Arabic that can be explained by the influence of Late Latin, including “vowel-length merger and the innovation of contrastive stress in archaic MA dialects … and the resulting extensive restructuring of morphological patterns that had previously relied on vowel length”, but the one phenomenon Heath discusses at length is the expression of possession through the preposition di. In Classical and Modern Standard Arabic, possession is expressed through the so-called “construct state” construction: the head noun (the thing possessed) appears in the special form, known as “construct state” (CS), whereas the complement (the possessor) is marked with the genitive case. (Note that the order is possessed-possessor.) Thus, in Classical Arabic, ‘a queen of a country’ is malikatu balad-in, literally ‘queen-CS country-GEN.INDEF’. Notably, the word in the construct state (e.g. malikatu in the above example) takes neither the definite article prefix al- nor the indefinite suffix -n: ‘a queen’ is malikatun rather than malikatu. Similar construct state constructions are found in some vernacular varieties of Arabic, particularly in Egyptian Arabic, where ‘a queen of a country’ is malikit balad. Although Egyptian Arabic (like most other vernacular varieties) has lost the genitive case marking, the first noun still appears in a special CS form, here malikit —compare to the non-CS form malika ‘a queen’.

In Moroccan Arabic, the construct state construction is used only in forming compound nouns; otherwise, it has been replaced by periphrastic prepositional construction, corresponding to ‘X of Y’, for alienable possession, and a double-marked possessive construction, corresponding to ‘Y’s X of Y’, for inalienably possessed nouns (kinship terms, body parts, etc.). Both constructions contain the possessive preposition ‘of’, realized as dyal or di or d-, depending on the context. (Hence, Heath refers to the possessive constructions formed with this preposition as D-possessives.) The latter (“double-marking”) construction can be exemplified by ḫa-ha [di l-mṛa] ‘her brother of the woman’.

Admittedly, the replacement of the construct state by periphrastic possessive constructions (containing a possessive preposition, typically refashioned out of nouns meaning ‘possession’, ‘property’, ‘thing’, ‘baggage’, or some such) occurred to some extent in other vernacular varieties of Arabic, including Egyptian, Levantine, and Iraqi Arabic, as well as in Hebrew. Thus, modern Hebrew features three possessive constructions, corresponding to the construct state in Classical Arabic and the two periphrastic possessive constructions found in Moroccan Arabic:

beyt ha-more ‘the teacher’s house’ (literally, ‘house.CS the-teacher’)

bait šel ha-more ‘a house of the teacher’ (literally, ‘house of the-teacher’)

beyt-o šel ha-more ‘the house of the teacher’ (literally, ‘house-his of the-teacher’)

What sets Moroccan Arabic apart from its Semitic siblings and cousins is the near-complete disappearance of the construct state (which is found only in compound formation), and the etymology of the possessive preposition di. According to Heath (2015: 16), “extant Maghrebi dialects divide into a type with genitive mtā’ (ntā’, tā’) from *mătā’- ‘thing(s), baggage’, and one with dī ~ d- ~ dyāl, rarely ddī. These D-possessives are dominant in all [Pre-Hilalian Moroccan Arabic] dialects (northern, … old urban dialects…), and are winning out over ntā’ in the MA koiné … To the east of Morocco, d- and/or dyāl occur spottily in archaic urban and Jewish dialects … on or near the Algerian coast… D-possessives are not reported for Tunisia, Libya or Malta”. In short, the possessive preposition di is limited to the areas where the influence of Late Latin is most likely to be found. Moreover, Heath (2015: 16-18) notes that “Arabists who have discussed Maghrebi D-possessives have looked diligently for a thoroughbred Arabic source but have struggled to find it”. He considers—and rejects!—several different Arabic-internal etymologies, including an etymological connection of the D-possessives to the Moroccan Arabic noun dyāl- ‘belongings, fortune’, as well as derivations from a demonstrative or relative morpheme.

Instead of looking for a “thoroughbred Arabic source”, Heath argues that the D-possessives derive from Late Latin preposition de. Because Arabic has had a three-vowel system, lacking an /e/, the preposition would be borrowed as di. Moreover, “the shorter form d-, proclitic to the following noun, may reflect an unstressed or elided [Late Latin] variant, or else attrition of earlier [Moroccan Arabic] di” (Heath 2015: 18), whereas dyal is analyzed by Heath as being “developed from the combination of the [Late Latin] version of Latin and inherited [Moroccan Arabic] dative l-, with some help from [Late Latin] pronominal prepositional phrases” (I refer the reader to Heath’s article for a more detailed discussion).

If Heath’s analysis is on the right track, it helps explain why Moroccan Arabic has some unique features, not found in Egyptian, Levantine, or Gulf Arabic.



Heath, Jeffrey (2015) D-possessives and the origins of Moroccan Arabic. Diachronica 32(1): 1–33.

McWhorter, John (2011) What Language Is. New York: Gotham Books.

Saussure, Ferdinand de (1959) Course in General Linguistics. New York: Philosophical Library.

Subscribe For Updates

We would love to have you back on Languages Of The World in the future. If you would like to receive updates of our newest posts, feel free to do so using any of your favorite methods below: