7 The Red Sea - Llacan

Transcription

7 The Red Sea - Llacan
à paraître. in Africa as a linguistic area, B. Heine and D. Nurse (eds). Cambridge University Press
Concours DR2 – N° 34/01
7 The Red Sea
Marie-Claude SIMEONE-SENELLE and Martine VANHOVE
(UMR 8135, CNRS, INALCO, University Paris 7)
1
Introduction
This chapter deals with the Semitic and Cushitic languages of an area which
encompasses both banks of the Red Sea in Africa and in Arabia, i.e. the coastal zone of four
countries: Sudan, Eritrea, Djibouti and Yemen. The reason for crossing the limits of the
African continent is twofold: the historical background, and the present socio-linguistic
situation, as we were able to understand it.
The geographic proximity of the Red Sea banks was a propitious situation for
population movement, even though navigation was quite difficult because of the wind system,
the coral reefs, the scarcity of shelters and of water supplies. Still, as early as the second
millennium BC, the African and Arabian populations were in permanent contact and the
region was already an important commercial route, also followed by migrants. Trading,
temporal or permanent trading posts, as well as military conquests favored the intense mixing
of populations.
A few well-known and significant examples may illustrate these contacts. The
archeological remains of Adulis harbor (today Zula in Eritrea, a few kilometers away from
Massawa, facing Hodeyda in Yemen) are evidences of the trade with southern Arabia since at
least the second millennium BC. Adulis was famous in all the ancient world1 as a very
important trading center for goods coming from inland Africa, specially Egypt and Sudan,
then from the high plateaus of Tigray, as well as from Arabia, Persia, and India. After the
decline of Adulis in the seventh century AD, the expansion of two islands further north,
Dahlak and Massawa, took over. Today, if Dahlak has totally collapsed, Massawa harbor is
booming again and is a favored place for contact with Yemen.
Military conquests and invasions also marked the history of the two regions. The Saba
kingdom played an important role in Aksum, then the Abyssinians occupied the western
slopes of the Yemeni mountains in the third century AD. In the sixth century AD the area
even became an Abyssinian protectorate. In the sixteenth century, the army of the imam
Ahmad ibn-Ibrahim al-Ghazi, better known as Gragn, i.e. ‘the left-handed’, the military leader
of Adal in Ethiopia, counted numerous Yemeni mercenaries from Zabid (Yemen). More
recently, in 1995, the Red Sea banks were again the center of a crisis with the claims of
Eritrea and Yemen over the Hanash islands.
Consequently, these exchanges and contacts had a deep influence on the culture of the
peoples of both banks, so much so that some talk of a Red Sea culture: “Evidently, there was
constant intercourse between the two sides of the Red Sea, and indeed a general “Red Sea
culture” is postulated” (Munro-Hay 2003:294). It has to be reminded here that the writing
system of Afro-Semitic languages such as Amharic and Tigrinya comes from the Ge’ez
syllabary which is derived from the South-Arabian alphabet used on the monument
engravings of the ancient kingdoms of southern Arabia.
Exchanges and contacts between human beings also take place through their
languages; this is why a study of the area of the two banks of the Red Sea from the viewpoint
of areal linguistics should be tested with reference to whether the Red Sea region can be
1
Cf. Pliny the Elder (1992).
understood to be a linguistic area. Such deep historical links from time immemorial until now
are bound to have produced some kind of linguistic effects.
As a matter of fact, language shift did occur. On the west bank of the Red Sea it is due
to the influence and through contact with the ancient South Arabian languages that the AfroSemitic languages spoken today in Eritrea and Ethiopia developed. They are genetically
directly related not to Arabic, but to the southern branch of the Semitic languages to which
the South Arabian languages belong. Afro-Semitic developed all along its history in contact
with Cushitic languages, but, on the east bank, ancient South Arabian was replaced by Arabic
from the seventh century AD onwards. Progressively, the South Arabian speakers were
confined in the Yemeni islands of Soqotra and Abd-al-Kuri, off Guardafi Cape, and in a
border zone between Yemen and Oman. Today’s modern South Arabian languages (MSA)
are, after an important evolution, the offsprings of the ancient languages.
As for the current sociolinguistic situation, we first have to point out that all the
peoples of the Arabian bank, as well as the great majority of the peoples of the African bank
are Muslims, so that religion is one of the reasons of the influence of Arabic on their
languages. Concerning education today, the languages taught at school vary according to the
different states. In Eritrea, there are three “working” languages, Tigrinya, Arabic and English,
and all mother tongues, among them Afar and Saho, are taught at primary level, except for
Dahalik speakers, whose language was discovered only in 1996, and for whom Arabic is the
medium of instruction. In Djibouti, Arabic is the official language, together which French,
which is the teaching medium, while Arabic is part of the curriculum. Afar and Somali, the
national languages, are not taught at school. In Sudan, the medium of instruction is Arabic,
and Beja, basically an unwritten language, has no official or formal status whatsoever.
In Djibouti, Arabic speakers of Yemeni origin are quite numerous in Djibouti city,
somewhat less in Obock and Dikhil and contacts are still intense with the areas of Khokha,
Taez, Aden and even Hadramawt. In addition, Afar speakers are in contact with Somali
speakers in the south, with Arabic in Obock, Tadjoura and Djibouti city, where a lot of
migrants from Somali and Ethiopia speak their own language (Amharic, Harari, Oromo, etc.).
In Eritrea, the Afar people dwelling along the coast are in contact with the Adeni and
Tihami varieties of Yemeni Arabic, those living in the north near Irafayle, in Bori Peninsula
are in contact with the Saho. Massawa is a real melting pot where Afar, Arabic, Saho, Tigre,
and Tigrinya are in contact. For Dahalik, contacts are mainly with Afar, Arabic, and formerly
with Tigre. For Saho speakers, contact languages vary according to the area: along the coast,
it is Afar, but in the Tigray region it is Tigre and Tigrinya.
In Sudan, most male Beja speakers are bilingual in Beja and Sudanese Arabic, but an
important proportion of females is still monolingual. In the south, the contact language is
Tigre. It must also be noted that there is an important settlement of Arabic speaking Yemenis
in Port Sudan.
In Arabia, in both Yemen and Oman, only standard Arabic is taught at school. Male
MSA speakers who live in the coastal villages or who are in contact with villages, are usually
bilingual in MSA and Arabic, but a lot of females are still monolingual. Some of the children
provided with schooling are bilingual. In Soqotra, contact with Arabic concerns mainly males
of the north coast and the inlanders trading with the capital city.
In spite of this well-known historical situation which goes on today along different
lines linked to social dynamics, studies on linguistic contacts between the two banks of the
Red Sea are just beginning. David Cohen, speaking from the Arabist viewpoint, was still
asking in 1994 (p. 18) whether it was:
2
compréhensible, pour ne prendre qu’un seul exemple, qu’une zone de brassage
immémorial comme la mer Rouge, où l’arabe a été confronté avec tant d’autres langues,
sémitiques, couchitiques, africaines, et même iraniennes et indonésiennes, reste fermée à
des recherches synthétiques sur ces contacts. (D. Cohen 1994:18)
From a general point of view, the question which must be considered is whether a
territorial continuity and linguistic contact necessarily imply the existence of a linguistic area
or Sprachbund in the classical sense of the term, such as defined e.g. by Emeneau (1980:124):
“an area which includes languages belonging to more than one family but showing traits in
common which are found not to belong to the other members of (at least) one of the families”,
or in looser manner by Campbell (1994:1471: “areal linguistics deals with the diffusion of
structural features across linguistic boundaries”), who specifies that the languages may be
“either unrelated or from different subgroups of a language family”.
However it is true that it is even more difficult to deal with this issue when the
languages concerned belong to the same phylum, as is the case with the Afroasiatic languages
we are dealing with. It may be difficult, or even impossible, to decide between genetic traits,
whether they are retentions or parallel innovations, and convergence phenomena. For some
linguists following Emeneau’s definition, genetic relation rules out the areal hypothesis, while
for others, such as Masica (1976) or Campbell (1994) it is a possibility providing that the
branches of the family are distantly related, as e.g. Cushitic and Semitic are.
When comparing linguistic traits in a contact situation, another problem is to decide
whether contact is unilateral, i.e. whether there is a dominant language from which other
languages borrow, or whether it is multilateral, without a dominating language, bringing
about convergence phenomena considered as characteristic of linguistic areas. For Tosco
(2000), unilateral contact is the rule in the Ethiopian zone such as characterized by Ferguson
(1976). But, following Sasse (1986) and Zaborski (1991), Sasse considers that it does not
exclude the possibility of smaller linguistic areas in the zone, such as an Eritrean-north
Ethiopian area.
One last problem, and not the least, concerns the choice of linguistic traits and the
difficulty of avoiding arbitrariness or ad hoc choice to characterize a linguistic area. The
items chosen by Ferguson (1976) were criticized by Tosco (2000) individually and generally,
questioning the validity of isogloss bundles for the hypothesized area. From a methodological
point of view we found it useful, in spite of Tosco’s grounded critics, to discuss a great deal
of Ferguson’s features in order to find out whether they could suit better the Red Sea zone,
and to enlarge them by a few others to test the areal hypothesis a bit further.
The Afroasiatic languages studied in this chapter all belong to the Semitic and
Cushitic families and more particularly to the following branches:
(i) Central Semitic: several varieties of Yemeni Arabic, mainly spoken in the Tihama
(the coastal zone) and in the mountainous Yafi‘ district, vehicular Arabic in Eritrea and
Djibouti;
(ii) Southern Semitic: Dahalik in Eritrea, an Afro-Semitic language discovered in
1996, spoken in the Dahlak archipelago, six modern South Arabian languages (MSA) spoken
in Yemen and Oman (Mehri, Soqotri, Hobyot, Bathari, Harsusi and Jibbali2);
(iii) Lowland East Cushitic: Afar (in Eritrea, Djibouti), and Saho (Eritrea);
(iv) Northern Cushitic: Beja spoken in Eastern Sudan. For this dozen of languages we
used the data collected during our recent fieldworks and the published sources available. For
2
For the last two languages, we relied on Johnstone’s data; see References.
3
the sake of a more complete account of language contact in this zone, we added data
published by linguists on two other Afro-Semitic languages, Tigre and, to as lesser extent,
Tigrinya, two languages which have been for a long time in contact with either Beja, Saho or
Dahalik.3
In what follows we are discussing phonetic, morphological and syntactic features, in
this order. We had to leave out lexical data for lack of enough relevant comparable material.
2
Phonetics
All the phonetic features noted by Ferguson (1976) as areal characteristics of the
Ethiopian plateau were strongly criticized by Tosco (2000) either because of their genetic
affiliation or their restriction to certain languages, typological naturalness or superficial
likeness.
We will not consider each of the eight phonetic traits Ferguson mentioned, but rather
discuss in detail the relevance, the diffusion and the distribution of four of them which could
be considered as important for the Red Sea zone, with the addition of three more which, at
first sight, seem characteristic. When possible, the phonetic features will be examined in
relation to the reorganization of the phonological systems due to common changes or to
linguistic contact. The aim is to try and find out if there are some phonetic features which
may be more specific to the zone. The phonetic traits discussed are the following: (pre)palatals and palatalization, “emphatics” (glottalization, pharyngealization, or retroflexion),
pharyngeals and laryngeals4, fricativization of stops and closure of fricatives, labialized velar
stops, and vowel length.
2.1 (Pre-)palatals and palatalization
Tosco (2000) criticizing Ferguson (1976), noted that the mere existence of fricative
(pre-)palatal consonants, so common worldwide, cannot be considered as an areal feature in
3
Except Tigre for which the data come from Raz (1983), Harsusi and Jibbali, all unreferenced data come from
our own fieldwork data and publications.
4
These three features correspond to features P2, P3, P4 and P5 of Ferguson.
4
itself. Still it is worth investigating in detail these phonemes as well as the often mentioned
palatalization phenomena in order to find out if there is any development particular to the
languages under consideration.
At the phonological level, when looking at individual languages, one finds important
differences concerning the number of phonemes, which range between one to eight, some of
them being very marginal, their distribution, and the voice and “emphatic” correlations.
Furthermore there is a clear-cut division between the two families of the phylum,
Cushitic and Semitic. Cushitic languages have the lowest number of palatal consonants: just a
single voiceless pre-palatal s in Afar, two voiceless s and ¸5 in Beja6, plus a marginal voiced
affricate ø, two in southern Saho with a voice correlation, s and z. The existence of six (pre)palatal consonants in northern Saho is due to the contact with Tigre and Tigrinya from which
the extra-phonemes, including the glottalized one, are most probably borrowed. So even
inside the Cushitic family, the three languages do not have the same systemic organization:
Afar (1): s
Beja (2): s, ¸ (ø)
southern Saho (2): s, z
northern Saho (6): s, z, s', ¸, G, ¢'
A few comments have to be added to the above list. The voiced palatals of Beja (ø)
and Saho (z) may alternate with the voiced stop d (or sometimes the retroflex ï in Beja),
depending on the dialect or even within the same dialect or idiolect in Beja. Morin (1995: 68)
notes that the existence of the Saho voiced z may be due to contacts with Tigre.
Semitic languages score a higher number of (pre-)palatals (between four to six), and
they have a more complex system in which the ‘emphasis’ and voice correlations are relevant:
Tigre (6): s, s', ¸, z, G, ¢'
Dahalik (4): s, ¸, z, G (+ marginal ¢' and S)
Arabic: (4 or 5): s, ¸, S, z (‚ or G)
MSA (5): s, ¸, ¸', z, G
From a historical point of view, MSA and Arabic  or  are the result of the
palatalization, in all phonetic contexts, of a Proto-Semitic stop *g which has been retained in
some Arabic dialects such as the Yafi‘i dialects, which have only four palatals. As for the
extra (and rare) glottalized phoneme ¸' of MSA it is either the result of the palatalization of
the velar k'7 or a particular evolution of s' (Simeone-Senelle 1997)
The “emphatic” correlation for (pre-)palatals (see also below) seems to be quite weak
in Dahalik: the post-glottalized ¢' is only found in two loans (so far) from Tigre, and the
pharyngealized S in eight words, most of them being borrowings from Arabic.
The historical processes of palatalization are still going on in a lot of the languages8 of
the area, but to various extents and concerning various phonemes. They are today either
5
The existence of ¸ in Beja may be the result of a contact-induced or internal development; see Tosco
(2000:338).
6
There is no “emphatic” palato-velar consonant in any variety of Beja, contrary to what Cohen (1988:250)
wrote, mislead by an erroneous reading of Hudson’s phonetic description and his changing notations.
7
Sometimes it is also a cognate of s'. For some words no phonetic explanation can be provided.
8
May be with the exception of Beja, if one does not count ø as a phonetic variant of d or ï.
5
phonetic variants or limited to certain morphemes, and they seem to be internal to each
individual language.
In several varieties of Afar spoken in Eritrea, the pre-palatal fricative s may
sporadically be pronounced as a palato-alveolar ~ in some words, usually when in contact
with a palatal phoneme (consonant or vowel), e.g. f`xr÷ ‘horn’ is f`x~÷ in Bori and Dahlak.
In Tigre, the palatalization of dentals and fricatives, widespread in Afro-Semitic but limited to
these languages, has become highly productive in the verbal morphology9 (Tosco 2000: 342).
It is the historical outcome of the loss of high vowels. Furthermore, ¸ and ¢ are also
allophones of z and t respectively when followed by the possessive suffix of the first person
singular -e (Raz 1983:4).
In MSA and in some varieties of Arabic, palatalization concerns different phonemes.
A change to a palatal ¸ occurred (in the vicinity of the high vowel i) for the voiceless velar k
in the feminine suffix of the perfect conjugation and the feminine possessive suffix: MSA
k™t™b-¯ ‘you (f.sg.) wrote’, amel- ‘you (f.sg.) did’.
In MSA, the change k > ¯ also occurred in the root of a few words such as ¯™b™dit
‘liver’; also for the emphatic k', which became ~' in some words: Soqotri ¯'édh™r ‘pot’.
In some Arabic varieties (the Hadramawt region in Yemen, Sudanese Arabic,
vehicular Arabic) and in Dahalik, etymological *g is usually strongly palatalized to Î or y (not
to  or ø) without any phonetic conditioning:
(vehicular Arabic): ÎdcHq` ‘island’, xdc, pl. xtcUc ‘grand father’, qŸx`≤`s ‘she came back’,
°◊x÷s ‘things’, xUx`≤ ‘he suffers’. (Dahalik): Îiran ‘neighbor’
In the Arabic variety of central Tihama in Yemen, it is the uvular *q which is
palatalized to Î: Îaada or Îiadeh ‘bed’.
Thus, it can be concluded that even if palatalization processes are certainly a common
feature in the Red Sea zone, they are by no means identical in each individual language, they
do not concern the same phonemes, and are far from being restricted to this zone, as it is also
found in the neighboring languages.
2.2 Glottalization, pharyngealization, retroflexion
Among the Afroasiatic consonants, there exists a genetic feature traditionally named
“emphasis”, a coarticulation which occurs in addition to the more common correlation
between voiced and voiceless consonants. “Emphasis” is a cover term for various phonetic
realizations: a post-glottalization in MSA, Afro-Semitic languages and some Cushitic
languages (outside the Red Sea zone), a pharyngealization in Arabic, and a retroflexion in
some Cushitic languages, among them Afar, Saho and Beja.10 What is at stake for the
discussion concerning a possible areal trait is precisely the distribution of the various
realizations of the emphatic correlation, which, as can be derived from the above list, is not
homogenous in the entire zone.
In Semitic, post-glottalization is not, synchronically, a feature common to all
languages of both banks of the Red Sea: it is absent in Arabic, and probably also in Dahalik,
where it occurs only in a few loans from Tigre (see above). As is well-known, speakers of
MSA languages from the southern part of the Arabian Peninsula were responsible for the
diffusion of Semitic in the west bank of the Red Sea. This is one of the reasons11 why it is
9
In the conjugation, the palatalization concerns also the lateral l (Tosco 2000:342).
There is also an implosive realization in some Cushitic languages.
11
In addition to phonetic reasons in evolution schemas (Haudricourt 1972, Hagège and Haudricourt 1978).
10
6
now generally admitted that glottalized consonants are original in both Afro-Semitic and
Proto-Semitic, and there is no need to postulate a Cushitic influence upon Afro-Semitic12. D.
Cohen (1988: 35) tentatively proposes a reconstruction with five ‘emphatic’ phonemes: *θ',
*t', *s', *λ', *k'.
Some glottalized proto-phonemes have developed into retroflex cognates. This is
Cohen’s reconstruction (1988: 250) for ‡ in Cushitic languages such as Afar, Saho and
Beja13. It most probably developed through a voicing of the consonant and a pre-glottal stage
(*t' > *'d). Other glottalized consonants seem to have fused with their non-glottalized
voiceless counterparts.
The retroflex articulation of the “emphatics” is very rare and marginal in Arabic. It is
only recorded so far for one phoneme in a north-western Yemeni dialect (Behnstedt 1987:
xxiv, 136), that of an-NaBìr. In two thirds of the vocabulary containing *D or *B, the
phonemes have fused in a pharyngealized inter-dental B, but in the rest, they are pronounced
as an affricate retroflex tÚ, without any phonetic conditioning. All other ‘emphatic’
consonants are voiceless and pharyngealized.
From the above review it is clear that even in the various outcomes of the emphasis, it
is not possible to find an areal feature of the Red Sea zone. Furthermore, according to
languages, it concerns a more or less important part of the consonantal system. In Cushitic it
is limited to the dentals, while in Semitic it also involves fricatives and velars.
2.3 Pharyngeals and laryngeals
Worldwide, pharyngeals and laryngeals are well-known for being particularly
predominant in Afroasiatic languages (Crystal 1987:165), but also for being subject to various
changes in language contact situations. Although these phonemes are a genetic feature, we
thought it worth investigating their present status in the languages of the Red Sea zone from
the point of view of possible areal developments.
The voiced pharyngeal fricative  is unstable in the Arabic variety of the coastal
Yemeni Tihama, where it is often weakened to a glottal stop ? (Behnstedt 1985: 43), but it is
still regularly preserved in Yafi‘i Arabic. In MSA  is also very unstable, even non existent in
some Mehri dialects and in Harsusi, where it often either becomes ?, lengthens the adjacent
vowel or changes its character. In all other MSA varieties,  is preserved. It is also well
preserved in all the Afro-Semitic languages and in Afar, but it is absent of Beja, where it may
have fused with ?.
As for the voiceless pharyngeal fricative H, it is stable in almost all the languages,
except Beja where it seems to have fused with the voiceless fricative laryngeal h.
Concerning laryngeals, the stop ? is usually weak in initial position where it is often
not pronounced, except in Soqotri (MSA) where it is stable, and in Mehri, Hobyot, and
Harsusi, where it occurs only in this position (but it is often strengthened to h or H). In other
positions, its preservation varies from one language to another: it is unstable in all positions in
Arabic, always pronounced between vowels and in final position in Dahalik (gaba?a ‘he
stayed’ gabbe? ‘he stays’), but often disappears in Tigre in final position. In Beja it is stable
12
For further details see Cohen (1988:13). One can still read in some recent works more careful statements (e.g.
Tosco 2000:342), or even contradictory ones (Thomason 2001:1642) with which we do not agree.
13
Somali, Sidamo and Oromo are also concerned.
7
in post-consonantal position but it does not occur in inter-vocalic or final positions. Afar and
Saho have no phonological ?.
The fricative h is present in all the languages of the zone, but in Dahalik it may
alternate with zero in initial position (halle or alle ‘to be’), and in some Beja varieties it
cannot occur between vowels.
Thus, it can be concluded that it is impossible to find a common pattern for any of the
pharyngeal and laryngeal phonemes.
2.4 Fricativization of stops and closure of fricatives
In some languages of both banks of the Red Sea zone, some obstruent consonants tend
to be weakened to fricatives.
In many Arabic dialects of Yemen (Tihama, Yafi‘) and in vehicular Arabic, the uvular
stop q can be fricativized to a velar [ö], a voiced uvular [$] or a voiceless [%], either as a free
or as a conditioned variant. A similar situation prevails in Dahalik for this consonant.
In Afar, where there is no uvular stop, it is the velar k which may be concerned in some
dialects or idiolects, but it is restricted to a particular phonetic context: k can be pronounced
as [x] in several varieties of Eritrea and in Djibouti when initial consonant in a cluster with a
final bi-labial m: yoxmeh ‘he ate’, laxmisáh ‘he is sick’. Such a process does not occur in
Saho, Beja, Tigre, Arabic or MSA.
But the reverse process is also possible, i.e. the closure of velar fricatives. This is the
case in Yafi‘i Arabic for the voiced velar fricative ö, which may become [q] (a free variant of
/ö/). In vehicular Arabic of Eastern Africa it can be pronounced either [q] or [g].
In Afro-Semitic and Cushitic languages which lack velar and uvular fricatives, the
process is also known in loans from Arabic. In Dahalik, *ö becomes [q] in initial position
only. Similarly, the voiceless velar fricative *x is always pronounced as a velar stop k14 in
Afar (koori < Arabic xºr ‘bay’), and in vehicular Arabic. But in Dahalik the etymological *x
of loan words is also changed to q (qalaa < Arabic xalaa ‘to make a mistake’).
On the whole, the two processes do not seem to be relevant for an areal diagnosis.
2.5 Labialized velar stops
Labialized velar stops kw and gw, which occur in parts of Cushitic (but not in Arabic
and MSA), are usually believed to be the source15 of these phonemes in Afro-Semitic. But
although they are quite common in the latter – they were already present in Ge’ez16 and are
systemic in northern Afro-Semitic Tigrinya – they are absent in both Tigre and Dahalik. They
are also missing in Cushitic Afar and Saho. Today, they are only found in Beja, in the most
northern part of the Red Sea zone. Their absence in almost all the languages of the west bank
of the Red Sea is interesting for the historical development of the languages of the area. The
fact that there are traces of the former labio-velars in Afar and Saho in labialized or velarized
phonetic realizations of some consonants in certain words (Cohen 1988:251), means that it is
possible that the two languages lost them under the influence of Semitic languages.
Considering the history of the Semiticization of north-eastern Africa, which started in the
north before proceeding further south, and the related increasing degree of Cushitic influence
from north to south (Tosco 2000:338), this is not altogether surprising. From a geographical
14
A general tendency in vehicular Arabic, also found in Juba Arabic for instance (Miller 1991:1071).
Mainly from Agaw, see Tosco (2000: 338).
16
A now extinct language, still in use in church liturgy.
15
8
point of view, this feature sets apart the west bank of the Red Sea from the Ethiopian plateau,
but, of course it is not an areal feature for the entire Red Sea zone.
2.6 Vowel length
The use of vowel length as a grammatical and lexical distinctive device is a genetic
feature. As it is well preserved in all the languages of the zone, except in Jibbali (MSA)
where it does not exist at all, and in Afro-Semitic where it is reduced to one vowel in Tigre (a
vs. Â), and probably also in Dahalik, it cannot be considered as a relevant feature for areality.
To sum up, it has not been possible to find a single phonemic or phonetic feature in
the Red Sea zone which is not genetically related or which could reflect a general
convergence tendency particular to the area. The sole contact induced phenomenon is an
influence of Semitic on Cushitic (i.e., the absence of the labio-velars).
3
Morphology
In this paragraph, we chose to enlarge the discussion about the morphological features
considered by Ferguson (1976), already criticized by Tosco (2000), to see if they could be
more specific of the Red Sea zone.
3.1 Lexical structure
Like most Afroasiatic languages, the languages of the Red Sea zone have a
conspicuous consonantal root. In Semitic languages, the structure of the lexicon, for both
verbs and nouns, is clearly organized in a root and pattern schema which determines semantic
and functional classes of words. Pattern morphemes constitute a close inventory consisting of
alternating vowels infixed inside the root, to which affixes can be added.
Such a structure is to a lesser degree also characteristic of the three Cushitic languages
of the area, as opposed to other Cushitic languages, in which the root and pattern schema is
extremely residual, and most often absent. In these languages, stem vowels are stable (D.
Cohen 1988: 256). Similarly, contrary to northern Afro-Semitic, in southern Afro-Semitic
languages, the system is archaic, as in Amharic (Leslau 2000: 40), and seems very marginal
in Gurage (Hetzron 1977: 53). The preservation of the root and pattern schema, although a
genetic feature, could be an areal feature differentiating the Red Sea zone from the rest of
Afro-Semitic and Cushitic. Still, in Africa, this structure concerns only a part, more or less
important, of the lexical organization. A much more common derivational device worldwide
is also in use, i.e. the addition of affixes to an invariable stem.
Anyhow, looking at the data in the languages of the Red Sea zone, it seems this
criterion is not as much ‘Semitic-biased’ as Tosco (2000:344) thought.
Hereafter is a discussion of the extension of the root and pattern schema in the
languages dealt with in this chapter.
3.1.1 Nominal patterns
The Semitic system being well-known, we refer the reader to Cantineau (1950) and
Cohen (1973-1979, 1988). In the following, we only give some examples in the less known
languages, MSA and Dahalik, a newly discovered language.
In MSA, it is used to derive different types of nouns from verbs, and vice versa, either
by vowel alternation or also with the addition of affixes: Mehri HºmÊl ‘to load’ or HÊmUl ‘to
carry’ > HÂmÊl ‘load’, mÊHmel ‘boat’; Jibbali raPá„ ‘to become soft’ > rÊPa„ ‘soft’.
9
In Dahalik, verbal nouns are derived from verbs either by a simple ablaut in the stem: sakaba
‘to sleep’ > sikáb ‘sleep (noun)’, or by ablaut and a stem prefix m-: teTÂra¸a ‘to sprinkle’ >
maTÂrí¸a. Nouns of instruments also follow this pattern: aHana ‘to grind’ > meHan
‘grindstone’.
In Cushitic, word formation with the root and pattern schema is often minimized
because it is used only in a small number of Cushitic languages, and it is less productive than
in Semitic: it concerns mainly the verbo-nominal derivation of the oldest type of verbs, those
conjugated with prefixes17. But there are also patterns particular to nouns of places and
instruments.
In Beja, nomen actionis of the prefix verb class are obtained either through a simple
ablaut: bir ‘to snatch’ > baar; kitim ‘to arrive’ > kituum, or also with the addition of the
prefix m-: dir ‘to kill’ > ma-dar ‘murder’, ma-door ‘murder(ous)’. The same complex
pattern is used for nouns of places and instruments (sometimes also with the lengthening of
the medial root consonant): ginif ‘to barrack’> mi-gnaf ‘camp’; himi ‘to cover’ > mehímmeey ‘blanket’; ?afi ‘to prevent, secure’ > m-?afay ‘nail, rivet, fastener’. Agent nouns
combine ablaut and the suffixation of a morpheme, -aana or -i: bir ‘to snatch’ > boor-aana;
kitim ‘to arrive’ > kaatm-aana or kaatim-i.
In Afar and Saho, where the verb class with prefixes always has an initial vowel in the
stem, nomen actionis are built through an ablaut to a vowel a: ookome ‘to arrive’ > áakam
‘preface, introduction’; ef¿e ‘to irrigate’ > af¿á ‘irrigation’; ubluge ‘to reach puberty’ >
ablaagé ‘puberty’. As in Beja, a prefix m- may be added to the ablaut (this pattern is often a
doublet of the preceding one): ef¿e ‘to irrigate’ > ma-fu¿(u) ‘irrigation’; ubluge ‘to reach
puberty’ > ma-blagá ‘puberty’. The same devices are used to form place and instrument
nouns: Afar u¿rufe ‘to rest’ > ma-¿rafa ‘place of rest’; u¿usure ‘to squeeze’ > ma-¿saará
‘juice squeezer’; Saho erde ‘to run’ > m-erdo (= arda) ‘running’. Another pattern for
nomen actionis is ablaut and a suffix -iyya (also often a doublet): ooboke ‘to be born’ >
aabuk-iyya (= ubka = m-aabaka) ‘birth’.
3.1.2 Gender
All languages of the Red Sea zone, like in Afroasiatic, have only two genders,
feminine (f) and masculine (m). As in many languages of the world, feminine is the marked
gender18. In the Semitic languages of the area and in Beja, nominal feminine is marked by a (v)t suffix. In addition to this suffix, in south Semitic mainly, another process is used, that of
a simple ablaut in the stem. This is true for a certain number of nouns, pronouns and
adjectives (also for some conjugations, see below, section 3.4.3):
Dahalik: nU¸ (m), nì¸ (f) ‘small, young’; Tawil (m), Tawal (f) ‘tall’.
Tigré (Raz 1983:33): qayÊH (m), qayÂH (f) ‘red’; sÊbur (m), sÊbÊr (f) ‘broken’.
Soqotri: het (m.), hit (f) ‘you (sg.)’; TáHrÊr (m), TáHrer (f) ‘wild goat’.19
17
Verbo-nominal derivation for the verb class with suffix conjugations uses suffixes added to an invariable stem
(that of the perfect tense): Afar wade ‘to testify’ > wad-eyna ‘witness’, or by reduction of the vowel length in
the stem ‡iine ‘to sleep’ > ‡in ‘sleep’.
18
A few nouns are intrinsically feminine, without a suffix.
19
The system seems to be still productive in Soqotri, but marginal in other MSA languages, eg. in Mehri: HE±Êr
(m), Hʱawr (f.) ‘green (pl.)’.
10
In Yafi‘i Arabic, the ablaut pattern for feminine (with a typical ) is residual in a few
nouns, adjectives, and in the demonstratives: ars? (m), ars (f),‘(bride)groom’; dn (m),
dn (f) ‘this’.
Afar and Saho have a feminine pattern distinct from both Semitic and Beja; feminine
nouns all have a stressed final vowel (á, ó, é, í, never u in Afar): awká ‘girl’, áwka ‘boy’.
3.1.3 Plural patterns
In all languages, two types of plural formation are predominant. The first one is
traditionally named external plural and consists of the addition of a suffix to the unmodified
stem. The second one, traditionally named broken plural, consists of a modified stem to which
a suffix or a prefix can be added.
All languages, except Beja, have an external plural with a -t suffix. Only Semitic
languages also have a nasal suffix (-m or -n), while only Cushitic languages also have a
vocalic suffix -a (also -wa in Afar and Saho). We will only discuss the so-called broken
plurals. Patterns of broken plurals are quite numerous in Semitic languages, but more
restricted in Cushitic.
All the languages use the simple ablaut as a device for broken plural, which often
entails a change in the syllabic structure of the word. Here are examples for some of the
different patterns:
Yafi‘i Arabic: ks, pl. ks ‘glass’; abl, pl. ibl ‘rope’; kurs, pl. kars ‘elbow’.
Mehri: HÊnìd, pl. HÊnºd ‘water-skin’; nÊKTÂt (f.),pl. nÊKÂT ‘dots’.
Jibbali: nid, pl. nud ‘water-skin’; nÊKTÓt, pl. nKOT ‘dots’; nibbÓt, pl. nbéb ‘bee’.
Hobyot: nObÉt, pl. nbéb ‘bee’.
Harsusi: nxÊrìr, pl. nxÊrºr ‘nose’.
Soqotri: náHrÊr, pl. náHrur ‘nose’.
Dahalik: ¿In, pl. ¿ayan ‘eye’; bettÂn, pl. betÂtin ‘sperm whale’; igir, pl. agar ‘leg’; kalºt,
pl. kalawit ‘kidney’.
Tigre: Ý™g™r, pl. Ý™gär ‘foot’; manSaf, pl. manÂSÊf ‘carpet’; qaSir, pl. qaSÂyÊr ‘(point of)
lance’.
Afar: Ha‡á, pl. Hoo‡ ‘tree’; arrabá, pl. arrób (arróbu) ‘tongue’.
Saho: anrab, pl. anrub ‘tongue’.
Beja: book, pl. bak ‘he-goat’; asil, pl. asuul ‘blister’; ganay, pl. ganeey ‘gazelle’; dhaar, pl.
dhar ‘landmark’; hasaal, pl. hasal ‘bridle’.
Broken plurals also combine ablaut with the use of affixes.
All the Semitic languages have a prefix for broken plurals which consists of a
laryngeal and a vowel, ?a-20 (also ha- or a- in MSA). This is not found in Cushitic.
Yafi‘i Arabic: qófl, pl. (a)-qfl ’bolt’.21
Hobyot: bìr, pl. ha-byºr ‘well’.
Harsusi: slEb, pl. hÊ-slºb ‘weapon’
Soqotri: ba¿áhar, pl. ?e-b¿ár ‘she-camel’.
20
Because of the weakness of the laryngeal stop in most languages, the prefix is often reduced to its vocalic
element whose timber may vary from one language to another.
21
In this Arabic variety, even the vowel can be dropped.
11
Dahalik: bEt, pl. a-bayit ‘house’; berk, pl. a-brak ‘knee’; gamal, pl. a-gmul ‘camel’, ibi¿,
pl. a-yÂbe¿ ‘finger’.
Tigre: naggal, pl. ?a-naggÊl ‘kid’; bÊggu¿, pl. ?a-bÂgÊ¿ ‘sheep’.
A nasal prefix m- is also used in MSA, but it is only very residual in Yafi‘i Arabic:
Mehri: nìdEx, pl. mÊ-nÂdÊx ‘smoke’.
Yafi‘i Arabic: nad, pl. ma-nwid ‘morning haze’.
As for suffixes of broken plurals, -(v)t is common to all Semitic languages:
Yafi‘i Arabic: isn, pl. in-eh22 ‘horse’.
Hobyot: KÀ±Êr, pl. KʱÓr-t ‘leopard’.
Tigré: masni, pl. masÂni-t ‘friend’.
Dahalik: HU, pl. Haw-Ât ‘brother’; ¿aqba, pl. ¿aqÂbu-t ‘tree’.
Tigre: ?arwe, pl. ?Âraw-it ‘snake’.
As for the -(v)n suffix of broken plurals, it only occurs in the Semitic languages of the
Arabian Peninsula (and in a few rare words in Dahalik: tºr, pl. ter-Ân ‘ox’, saqar, saqr-Ân
‘turtle’):
Yafi‘i Arabic: dd, pl. dd-n ‘worm’.
Jibbali: gífún, pl. gÊfÊn-ín ‘tulchan’.
Hobyot: KIrÓS, pl. KirS-ìn ‘mosquito’.
In all the southern Semitic languages, broken plurals may combine prefix + ablaut +
suffix:
Mehri: bºb, pl. Ha-bwEb-Êt ‘door’; Hirì-t, pl. a-Hyár-t-Ên ‘she-donkey’.
Dahalik: HÂn, pl. ?a-Ha?an-et ‘boy, child’23.
Tigre: GÊna, pl. ?a-Gann-it ‘child’.
None of the Cushitic languages has prefixes for broken plurals, but the suffix -a can be
used in addition to the ablaut.
Afar: ¿alé, pl. ¿aleel-á ‘mountain’; garbó, pl. garboob-á ‘forest’.
Saho: akat, akt-a ‘rope’.
Beja: dhaar, pl. dhár-a ‘landmark’; hasaal, pl. hasál-a ‘bridle’.
Beja is the only language where the position of stress can be distinctive for the
expression of plural forms: ha‡áa-b, pl. há‡aa-b ‘a lion (A)’ [WhAT IS “A”?]. This device
is known in other Cushitic languages (Zaborski 1986).
In addition to the above device, there is also in all the languages, except MSA, a plural
pattern by reduplication of a root consonant. Although reduplication patterns are quite
common (to various extent, see Zaborski 1986: 293) in Cushitic (as internal gemination in
Semitic) it is noteworthy that ablaut is also involved in this kind of patterns in all the
languages of the zone (more or less according to the dialects), but not elsewhere. The
reduplicated consonant differs from one language to another and the structures are different.
In Semitic, an inner consonant is doubled, in Afar and Saho, the root-final consonant is
reduplicated (with the addition in Afar, of a vocalic suffix and the lengthening of the final
vowel of the singular form), and in Beja, where there are only traces of the morphological
22
In most Arabic varieties, the final -t has become -h or zero in pausal form, but it is still -t in the genitive
construction of the noun and with possessive suffixes.
23
So far, it is the sole example provided by the data. It has also a variant without a prefix: Ha?an-et.
12
device in a small number of adjectives (Reinisch 1893:§119 and Zaborski 1986:4), it is the
initial consonant which is reduplicated, always followed by a long vowel aa:
Yafi‘i Arabic: agz, pl. aggez ‘old person’.
Tigre: ¸Êngul, pl. ¸anaggÊl ‘adult’.
Afar: garbó, pl. garboob-á ‘forest’; saró, pl. saroorí ‘cloth’; iná, pl. inaaní ‘mother’.
Saho: ¿okka, pl. ¿okkak ‘ear’, gaba, pl. gábob ‘hand’.
Beja: win, pl. waawin ‘big’.
Even though broken plurals can be considered as a genetic feature, they are,
nevertheless, characteristic of the entire Red Sea zone, and the fact that they are very rare or
absent in other Cushitic languages (Zaborski 1986) speaks for a possible influence of Semitic
upon the Cushitic languages of the area.
3.2 Case systems
There is a clear-cut genetic division between Cushitic and Semitic as far as case
systems are concerned: present in the former, but absent in the latter, even though Arabic had
once a nominal case system still preserved in the written variety.
In Afar and Saho, the system is asymmetric and cases are only marked for the category
of masculine nouns. In Afar only, those ending with a consonant have three cases, two of
which are marked: oblique has a final vowel in harmony with the last vowel of the stem, and
genitive has a final stressed vowel -í24 (before a post-position only); the unmarked form is
used for both nominative and accusative:
¿afár af-i-h agaarada-h (Afar language-G-POST watch-for) ‘for the Afar language watch’;
kabel ‡aamite (sandals I bought) ‘I bought sandals’; dukkáan num (shop man)
‘shopkeeper’; furum fa‡a (bread I.want) ‘I want bread’; furum me¿e ‘bread is good’;
me¿e furuumi-h suruy (good bread.G.POST smell) ‘the smell of good bread’.
In both Afar and Saho, masculine nouns ending with a vowel have only two cases
which are marked by the ablaut of the final vowel. The marked form ends with an í and
expresses both nominative and genitive, the unmarked form is for the so-called absolutive
(accusative and oblique): áwka uble ‘I saw the boy’, áwka-h taHée ‘she gives to the boy’,
awkí ge‡e ‘the boy went away’, idaltí ba‡á ‘the old man’s daughter’ (idáltu ‘old man’).
The case system of Beja is different and less restricted than that of Afar and Saho. For
masculine and feminine nouns whose syllabic structure is either monosyllabic or bisyllabic
(with a first short vowel other than a) there is an opposition between three cases, nominative,
accusative and genitive, which is marked by the ablaut of the prefixed article (Vanhove and
Hamid Ahmed, in press).
sg.
pl.
N (t)uu- (t)aaA (t)oo- (t)eeG (t)i(t)i24
In Afar the genitive marker is -ti in monosyllabic masculine nouns: Han-tí suruy ‘smell of milk’ (Hán(a)
‘milk’). This marker is also used in Saho.
13
Table 1. Beja: Case marking in the article
For indeterminate nouns, only two cases are marked, the third one, the nominative, is
unmarked. The accusative is marked by a final consonant -t for all feminine nouns, and -b for
masculine nouns ending in a vowel. The genitive, similarly to the article, is characterized by a
vowel -i (-ee for plural forms) suffixed to the head noun.
If case system is ruled out as a possible areal feature, it has to be pointed out that the
patterns used in the Cushitic languages of the zone do not conform to the predominant one
typical of North African languages. Heine and Leyew (this volume, chapter 2) note that in this
area it is the nominative case which is marked, and the accusative unmarked, but if Afar and
Saho follow more or less this pattern, indefinite nouns in Beja present a reverse pattern.
3.3 Personal pronouns and possessives
We are only discussing below the object and possessive pronouns because they are
less subject to innovations than subject pronouns, and thus they can be more easily compared.
Below is a chart for object and possessive pronouns in all the Semitic languages25:
1sg
2m
2f
3m
3f
1pl
2m
2f
3m
3f
Mehri
-i / -
-k
-
-h
-s
-n
-km
-kn
-hm
-sn
MSA
Soqotri
-yh / -
-k
-
-h, -
-s
-n
-kn
Afro-Semitic
Arabic
Dahalik
Tigre
Yafi‘ Tihama+ V
-(h)e / -ni, -he -ye / -ni - / -ni  / -ni
-ka / -ak
-ka
-vk
-k
-ki / -ik
-ki
-v
-k
-(h)u / -o, -u
-(h)u
-ah, -u -o(h), -ho
-(h)a
-(h)a
-h
-h
-na
-na
-na
-n
-kum
-kum
-kum
-kum
-kan
-kn
-kn
-hn, -n -(h)um, hun
-(h)om -hum
-hum, -hun
-sn
-(h)an
-(h)an
-hn
Table 2. Object and possessive pronouns in Semitic
In all the Semitic languages, object pronouns and possessives are identical, except for
the first person singular. In Dahalik there is in addition a slight difference in the syllabic
structure for the second person singular, but the base is the same, and the difference in the
third person masculine singular is due to the weakness of the laryngeal h.
In both vehicular Arabic and Tihami Arabic, contrary to the other Semitic languages,
there is no opposition of gender for all the plural forms (like in Cushitic languages, see
below), and in the singular it is lost also for the second person. The whole system is similar to
that of Afar (see below), the contact language for vehicular Arabic. But the gender distinction
tends to be lost in several Arabic varieties, like in the dialect of Aden.
In Soqotri, the absence of gender opposition in the plural is limited to the second
person plural.
25
When possessives and object pronouns are different, possessives are to the left of the oblique bar and object
pronouns to its right. A coma indicates a variant.
14
In MSA languages26, the pronouns with a palatal fricative are cognates of an old stage
in Semitic, lost in Arabic.
The languages of the two branches of Cushitic of our sample have different systems.
In Beja, like in the above Semitic languages, object pronouns and possessives are both
suffixes, and quite similar in shape, but possessive modifiers differ according to the case of
the head noun, a difference marked in the vowel of the suffix.
object
possessives
+ prf & ipf + durative + N. sg. nom. + N. pl. nom. + N. sg. acc.
-i (-u)
-i
-aa
1 sg. -heeb
-i
-hook
-uuk
-uuk
-aak
-ook
2
∅
∅ (-hos)
-uu
-aa
-oo
3
-uun
-uun
-aan
1 pl. -hoon
-oon
-hookna
-uukna
-uukna
-aakna
-ookna
2
-uu
-aa
-oo
∅ (-hosna) ∅
3
Table 3. Beja object pronouns and possessives
+ N. pl. acc.
-ee
-eek
-ee
-een
-eekna
-ee
-i (or -u, a dialectal variant), the cognate of the Afroasiatic bound form for the object
pronoun of the first person singular, is only used after the durative, which is an archaic form,
but the object pronoun -heeb used after the perfect and the imperfect is an innovation. There
are no object pronouns for third persons except in the northern dialect of Bishariyin, as
described by Reinisch (1893). It is also possible to express cross-referencing of the addressee
by adding a gender marker after the bound forms, -a for males, and -i for females.
If the head noun is a genitive form, it keeps the -i / -ee marker before the possessives.
Feminine nouns with possessives always show the feminine suffix -t, but gender is unmarked
for masculine nouns.
Although they are similar in form, Afar and Saho have two distinct systems for object
pronouns and possessives: object pronouns are always free pronouns, and possessives are
bound forms but prefixed, contrary to all other languages.
object pronouns
possessives
sg. 1 xË'n(
pl. m·'d(
sg. x‚,
pl. m‚,
2m jË'n(
r‚m+r‚hmh

jÔ, 
r‚m,
3m j÷'`(
j·m+j·dmh 
j÷x, 
j·m,
3f s·s+s·dsh
s·s,
Table 4. Afar, Object pronouns and possessives
Comparing the various systems of all the languages of the Red Sea zone, one notes a
tendency towards a reduction of gender opposition, Semitic languages being usually more
conservative, and particularly the Afro-Semitic languages of the area (contrary to other AfroSemitic ones such as Amharic, where it exists only for the singular; cf. Leslau 2000: 19). Afar
does not even show a gender opposition in the plural forms, still present in Beja and in the
conservative varieties of Arabic.
From a genetic point of view, first and second persons are common to all the
languages, and all Semitic languages have similar systems27. But for third persons, Cushitic is
26
27
Dual forms for 1st , 2nd and 3rd persons are not given because they are not attested in other languages.
The correspondence h/¸/s is regular between MSA and within Semitic.
15
distinct from Semitic, and moreover Afar differs from Beja, with third person possessives
with a velar consonant k.
From the above description it seems that the pronominal systems do not display any typical
areal features; rather, they are the result of genetic affiliation. Innovations are particular to
each language.
3.4 Typology of verbal systems
All the languages have a fundamental aspectual opposition between a perfect and an
imperfect which is morphologically characterized by:
(a) the inclusion of personal (PI) and number (NI) indices in the verb form (so that overt
subject noun phrases are not necessarily required),
(b) two types of conjugations: one with prefixes, one with suffixes,
(c) ablaut,
(d) derivation with affixes,
(e) the development of the basic system with auxiliaries.
3.4.1 Prefix conjugation
The prefix conjugation, used in both Semitic and Cushitic languages of the Red Sea
zone, is known to be the oldest one for both families. The sequencing of morphemes is always
the following:
PI + verbal stem28 + NI (with gender opposition for second persons).29
But its function differs in each family: in Semitic languages the prefix conjugation has
an aspectual value of imperfect while in Cushitic it marks a class of verbs as opposed to
another one. In Beja, verbs with a suffix conjugation represent almost 60%, and more than
one third in Afar and Saho (Cohen 1988: 256). It is noteworthy that this archaic verb class is
only represented today in five verbs in Somali and South Agaw and is totally absent in other
Cushitic languages. It is possible that contact with Arabic, and previously with Afro-Semitic
in Eritrea, helped to preserve this pattern in the Cushitic languages of the zone.
3.4.2 Suffix conjugation
The suffix conjugation is used in both Semitic and Cushitic languages of the Red Sea
zone, but again it does not have the same value in both families. In Semitic, it is a perfect
aspect, but in Cushitic it is one of the two morphological classes of verbs, the most recently
created one. Furthermore, even if there are common traits, these conjugations do not share the
same origin. Each family used different grammaticalization schemas and processes, based on
either the nominal sentence or on periphrastic constructions with auxiliary verbs (Cohen
1984).
In Arabic for instance, the actual suffix conjugation is a recent innovation where
former personal pronouns fused with nominal stems. In Cushitic, the PIs are the result of the
grammaticalization of an auxiliary verb meaning ‘say’ or ‘be’, conjugated with prefixes
(Cohen 1984, 1989, Zaborski 1975). Thus, the suffix conjugation in itself is not an areal
feature, neither formally nor functionally, but the result of independent innovations in each
family.
28
29
In some dialects of MSA, gender opposition is marked in the imperfect by an ablaut in the verbal stem.
Except in Afar and Saho where gender opposition is marked in 3sg only.
16
3.4.3 Ablaut
The ablaut of the stem is generally considered as characteristic of a large part of
Afroasiatic phylum (Greenberg 1955a quoted in Hayward 2002 [2000]: 92). We examined
above (section 3.1) its role in the nominal morphology, but it is also an important
morphological device in the verbal system. Their morphological and semantic roles vary
according to languages and can be relevant for diatheses, gender opposition, aspectual
opposition and semantic derivation. Below we discuss the various functions in each language.
3.4.3.1 Diatheses
In the Arabic varieties, the ablaut as a morphological device in the active - passive
opposition is quite residual. Behnstedt (1987: 19, 21) recorded some traces in a few north
Yemeni dialects, and the Arabic of Yafi‘ still has a marginal use (murn ‘they were caught
in the rain’ (maar ‘it rained’), nuul ‘it was taken off’ (naal ‘he took off’), yuql ‘it is
said’ (yiql ‘he says’). This type of ablaut is not recorded in Tihama except in very rare
fossilized forms like yuqÂl ‘it is said’.
On the contrary, in all MSA, ablaut is still a device for the opposition between active,
middle and passive voices:
Mehri lÊbUd ‘to hit’, passive: lÊbEd ; FÊbºr ‘to break’, middle: FìbÊr
Soqotri ¸ŸdÊ ‘to divide’, passive ¸ìdÊ.
But in Afro-Semitic ablaut is not used for diathesis. For Cushitic the ablaut for
diathesis is only found in Beja, but only in the perfect of the prefix class of verbs. There is an
opposition between an active and a middle-reflexive voice characterized by a vowel a: fif ‘to
pour’, faf ‘to be poured’; rimid ‘to avenge’, ramad ‘to take revenge’.
This short review shows that the use of ablaut for voice distinction is used in both
banks of the Red Sea and in both language families, but it is today too restricted, and at times
marginal, to be considered in itself as an areal feature.
3.4.3.2 Mood
The MSA are the only Semitic languages to make use of the ablaut to form moods.
The opposition between indicative and subjunctive is limited to certain conjugations only and
concerns the stem: Mehri indicative imperfective yÊ-dkºk ‘he springs at’, subjunctive lÊdkEk or yÊ-dkEk.
In Afar and Saho the opposition is marked in the ablaut of the inflectional morphemes:
Afar (imperfective) y-amaat-e, (subjunctive) y-amaat-u ‘to come’; Saho (imperfective) rabÀ, (subjunctive) rab-ó ‘to die’.
Mood formed by means of ablaut is unknown in all other languages, but again both families
are concerned.
3.4.3.3 Aspect
In the Semitic languages, ablaut is used for aspect, usually alongside other
morphological devices, to mark the distinction between perfect and imperfect:
Arabic qÂl / yiqUl ‘to say’, nazal / yinzil ‘to go down’.
Mehri FÊbºr / yÊFºbÊr ‘to break’, middle FìbÊr / yÊFbºr.
Dahalik nasa?,` . nasse? ‘to take’; na¸¸af-a . na¸¸if ‘to dry’; dárbe / dárbi ‘to throw’;
sé¿e / sá¿e ‘to hurry’.
17
Tigre kÂtab-a / lÊ-kÂtÊb ‘to vaccinate’.
The situation is somewhat similar in Cushitic languages. In all three of them the ablaut
occurs in the stem for the old prefix conjugation, but in the suffix conjugation in Afar and
Saho, it only concerns the inflectional morphemes:
Afar: yen / yan ‘to be’; emeete / amaate ‘to come’; fa‡-e / fa‡-a ‘to want.’
Saho: yine / yane ‘to be’; rab-À / rab-é ‘to die’.
Beja: i-dif ‘he went’, in-diif ‘he goes’; i-ktim ‘he arrived, kantiim ‘he arrives’, i-ktim-na
‘they arrived’, ee-kátim-na ‘they arrive’.
The use of ablaut for aspectual distinction is the sole usage which concerns all the
languages of the zone.
3.4.3.4 Gender
The use of ablaut in the stem for the expression of gender opposition is restricted to
the MSA languages, and only for certain conjugations:
Mehri ktéb (m), ktíb (f) ‘write!’; tÊFºbÊr (m, sg.), tÊFìbÊr (f, sg.) ‘you break’.
Soqotri tÊKÓfÊd (m, sg.), tÊKÓfid (f, sg.) ‘you go down’.
3.4.4 Semantic derivation
The ablaut is a device for semantic derivation in both families and on both banks of
the Red Sea, but it is limited to two branches.
Among Semitic languages, only the MSA use ablaut as a derivation device for the
intensive, causative or transitive values: Mehri rEkÊb ‘to ride’, (a)rºkÊb ‘to put (a pot) on
the fire’.
As for Cushitic, Beja is the only language to use ablaut as a derivational device for the
expression of intensive (generally for transitive verbs with a plural object), but only for the
prefix class of verbs: bis, INT. bos ‘to bury’; dir, INT. daar ‘to kill’; birir, INT. baarir ‘to
spill’.
Even though it is restricted in some of its usages, the use of ablaut as a productive
morphological device with various functions, both in the nominal and the verbal systems,
concerns both families of the Red Sea zone. It is even more remarkable if one considers that it
is by far less productive (and even sometimes absent) in other Afro-Semitic and Cushitic
languages.
3.4.5 Verbal derivation
Apart from the ablaut device, there are several derivational morphemes with similar
values which are common to all or most of the Semitic and Cushitic languages: a sibilant s or
¸, a dental t, a nasal n or m. The laryngeal ? or h is an innovation in Semitic only, and y or w
an innovation in Afar only. In Semitic, the derivational morphemes are either prefixed or
infixed, while in Cushitic they are prefixed or suffixed according to the verb class, sometimes
circumfixed for some verbs in Afar and Saho. So the Red Sea zone does not display the
preference for derivational suffixes as many African languages do (Heine and Leyew, this
volume). Derivational morphemes are not a good diagnosis feature for areality as they are
either genetically inherited or restricted innovations.
18
3.4.6 Consonantal gemination and reduplication
The criteria of reduplicated intensives outlined by Ferguson (as a verbal and adjectival
device) is considered by Tosco (2000: 349), for Afro-Semitic, as induced by the contact with
Cushitic. It consists of the reduplication of at least one syllable, sometimes of the whole stem.
For semantic and formal reasons, it is worth comparing it to the gemination of the inner-root
consonant in Semitic languages. It occurs in Arabic and in Afro-Semitic, but not in MSA,
even though the former two are directly related to the latter languages.
In Arabic, the gemination of the second root consonant is a derivational device which
changes the valency of the verb or forms denominative verbs. More important for the purpose
of comparison, it can also be used to form intensive verbs, just as the reduplication device of
Cushitic and Afro-Semitic: kasar ‘to cut’, kassar ‘to cut into pieces’. Unlike Arabic,
gemination in Dahalik and in Tigre is now an aspectual device, like in most Afro-Semitic
languages. In these two languages it forms the imperfect, but it is known that it was originally
an intensive derivational device (cf. e.g. Leslau 1956: 93).
The gemination device is not used in Cushitic, but languages have recourse to total or
partial reduplication (initial or final, or even medial in some Beja verbs) of the verbal stem in
order to derive intensive verbs (augmentative or attenuative): Beja hireer ‘to walk’,
hireerhireer ‘to walk hastily’; tam ‘to eat’, tamtam ‘to gobble hastily’, tatam ‘to eat
slowly’; kitim ‘to arrive’, kÂtatim ‘to arrive at many places’. Afar walal-e ‘to discuss’,
walamwalal-e30 ‘to chat’; ge‡-e ‘to go’, ge‡amge‡-e ‘to wander’.
A similar process is also used in Afro-Semitic, in both Tigre and Dahalik, but partial
reduplication concerns the internal syllable. It is quite productive to derive intensive verbs,
but in Tigre it only concerns tri-consonantal roots: Dahalik bataqa ‘to cut’, batataqa ‘to cut
into pieces’; Tigre: kal?a ‘to pass time’, kalÂlÊ?a ‘to pass a short time’.
In this instance again, it is difficult to conclude in favor of an areal feature.
Reduplication for the expression of intensity is quite a common device cross-linguistically
worldwide and reduplication as such is a case of unilateral influence from the Cushitic
substrate upon Afro-Semitic.
4
Syntax
In the discussion below, we group together several of the grammatical features
outlined by Ferguson which belong to the domain of syntax.
4.1 Constituent order
In the Red Sea zone, verb final order, a quite common order worldwide but not in
Africa (Heine 1976), is restricted to the African bank. It is not found in the Arabian Peninsula,
where VSO order is predominant (or, less frequently, SVO). Still, it is noteworthy that in the
vehicular variety of Arabic of north-east Africa, the predominant constituent order in the
sentence is SOV, the contact with Cushitic and Afro-Semitic languages being responsible for
this unusual order in Arabic. At any rate, considering this clear-cut division between Africa
and Arabia, and the extension of this feature to other languages of north-east Africa, SOV
order cannot be considered as an areal feature of the Red Sea zone.
Nor is the typologically related order of clauses in complex utterances a common
feature of the zone: dependent clause - matrix clause is the canonical order in Africa
30
In Afar the linker -m- is obligatory between the two parts of the reduplicated stem.
19
(although not always exclusive), but it is the reverse in Arabia for most adverbial and
complement clauses, except for temporal and hypothetical clauses which most often precede
the matrix clause.
The word order for adnominal modification is more complex and depends on the word
category and the semantic types of relations. Once again, it is not common to the entire
geographical zone, and for each type of modifiers, the order is not always consistent with the
basic order of the major constituents of the clause.
In north-east Africa, adnominal modifiers often precede the head noun. This is true for
referential modifiers such as the article, where it exists31, for the demonstratives32 (except in
Dahalik where they follow the head: ediret-da (island-this) ‘this island’), and for
quantitative modifiers33, including numerals.
But the position of locative and qualitative modifiers is more complex: it varies
according to languages, and also language internally, according to different rules.
In Tigre and Beja, qualitative modifiers may precede or follow the head noun. In Beja,
adjectives normally precede indefinite nouns but follow definite ones, but in Tigre there is no
rule according to degrees of determination, and both orders are possible, although adjectives
preceding the head noun seem to be more frequent (sannet wart (good work) ‘good work’
but wddy sannet (deed good) ‘good deeds’). In Afar, adjectives, a rare category, precede
the head noun (mee awká (good / girl) ‘a good girl’), but in Dahalik they follow it
(may wauy (water hot) ‘hot water’).
Relative clauses, another category of adnominal qualitative modifiers, also exhibit
diversity. If, like many languages of the world, all the languages of the area are head external,
there is no unity in the sequencing of the different parts of the utterance. In all the Semitic
languages of Arabia and in Cushitic Beja, the order is head noun - relative clause - matrix
clause. This is also one of the two orders possible in Afro-Semitic Dahalik, but the second
order type, head - matrix clause - relative clause, is quite unusual in this part of the world and
particular to this language:
arba gidda tuwaddiy- di muxla inde xabaatt-a
on
she cooks-3F.SG
four shares she does-3F.SG REL plate
‘She divides it into four portions which she cooks on the plate.’
In the other African languages, Tigre, Afar and Saho, the relative clause precedes the
head noun:
Afar narúw taaïigen
numu-k
ïaamite.
ram
know2PL.IPF man-from buy1SG.PF
‘I bought a ram from the man whom you know.’
In Afro-Semitic, this constituent order is considered as a typical Cushitic influence
(Thomason 2001: 1642), but, as we saw above, it is not shared by all the Cushitic and AfroSemitic languages of the area. A possible explanation of the fact that the head noun precedes
31
Afar has no article.
Demonstratives co-occur with the article when there is one, except for Dahalik.
33
In Beja the position of quantitative modifiers depends on the modifier itself: kass- ‘all’ may precede or follow
the head, weer ‘other’ precedes it, but raw ‘other’ follows, while -ka ‘each’ is an invariable suffix. In Tigre
both positions are possible.
32
20
the relative clause in Cushitic Beja could be a stronger Semitic influence (probably from
Arabic) on this language.
In Afar, the reverse order is also possible with the head preceding the relative clause,
the difference being a lower degree of determination:
narúw
taaïigen numu-k
ïaamite.
ram
know2PL.IPF
man-from
buy1SG.PF
‘I bought a ram from a man whom you know’.
As for locative modifiers, in Tigre they always follow, whether they are full nouns
(walat n™gus (daughter king) ‘the king’s daughter’) or possessive pronouns (see above,
section 3.3), but Afar and Saho have the reverse order and all types of locative modifiers, be
they full nouns or possessives, precede the head (¿eel-í lee (well-G water) ‘water from the
well’). As for Beja and Dahalik, they have asymmetric patterns, but different ones: in Beja
locative modifiers precede when they are full nouns (tak-i kam (man-G camel) ‘the man’s
camel), but follow when they are possessives; in Dahalik possessives always follow the head,
but the position of full nouns varies according to the type of construction: the modifier
follows the head in the synthetic construction (aw-walet (father-daughter) ‘the daughter’s
father’), but precedes it in the analytic construction (walet na amad (daughter of Hamad)
‘Hamad’s daughter’).
In the Arabian Peninsula, both in Arabic and MSA, it is the head noun which most
often precedes the modifiers, except for the article, where it exists, the possessive
construction, and numerals (other quantitative modifiers may precede or follow).
The main orders for modifiers vis-à-vis the head noun (H) are summed up in table 5. It
shows that Afar and Saho are the sole languages to be consistently head-final (if one excludes
the possibility of the reverse order for relative clauses in Afar).
Basic order
Article
Demonstratives
Numerals
Adjectives
Relative clause
Genitive
Possessives
Beja
Afar
Saho Tigre
Dahalik
SOV
SOV
SOV SOV
SOV
H2
H2
H2
H2
H2
H2
H2
H1
H2
H2
H2
H2
H2
H2, H1 H2
H2
H2, H1 H1
H1
H2, H1 H2
H2
H1
H2
H2
H2
H1
H2, H1
H1
H2
H2
H1
H1
Table 5. Position of the head noun vs modifiers
MSA
VSO
H2
H1
H2
H1
H1
H2
H1
Arabic
VSO
H2
H1
H2
H1
H1
H2
H1
As for verbal modifiers, their position is more stable and differs on geographic
grounds: the verbal modifier, i.e. the auxiliary verb, follows the main verb in Africa,
irrespective of genetic affiliation, while it precedes it in Arabia. This is consistent with the
respective basic word order.
On the other hand, linkers display a clear-cut division between genetic families,
irrespective of geographic distribution. All Cushitic languages have postpositions and postposed conjunctions, but Afro-Semitic languages, although SOV languages, have prepositions
and pre-posed conjunctions. Dahalik is particular in that it is the sole language in which the
21
place of subordinating conjunctions may vary, either at the end of the adverbial clause or in
front of the verb, depending on the linker:
ba?aHa-k-ka mingabbi (arrive-3M.SG.PF-you(SG) if) ‘if it reaches you’
Humum nahar
elliko
i-ma?ì-ni
madrase.
sick
when
have1SG.PF NEG-go1SG.IPF-NEG school
‘When I am sick, I do not go to school.’
ana
asmara muqo bee
I
Asmara when
I arrive
‘When I arrive in Asmara’
In addition to the fact that there is no unity between the two banks of the Red Sea, it is
noteworthy that there are also discrepancies between the different correlated patterns
concerning constituent order, language internally as well as cross-linguistically, as already
noted by Tosco (2000: 344) for the Ethiopian plateau. It seems that the influence of Cushitic
languages on Afro-Semitic languages has been less important on linkers than on other
constituent orders, which is in line with the fact that complementizers do not fit easily in the
pattern of head-final languages (they are ‘bad patterners’, as Primus (2001:856b) puts it). The
entire system of constituent order may reflect also a contact-induced chain reaction in delayed
typological effects, an explanatory pattern proposed by typologists (see e.g. Thomason 2001:
1644).
From the great diversity described above it can be deducted that the constituent order
is not a criterion which is relevant for the Red Sea zone. Still, there is a clear division between
the two banks of the Red Sea, with Tigre tending towards the African bank and Dahalik
towards the Arabian bank, as table 5 above shows.
4.2 Quoting clauses (reported speech or discourse)
Ferguson (1976:71) mentioned direct reported discourse and its use, in more or less
grammaticalized constructions, to express various epistemic modalities of intention and goal
as a characteristic feature of the Ethiopian plateau. Tosco (2000: 345) also discusses this, as
well as its grammaticalized outcomes, as “a good example of Cushitic influence upon
Semitic”. Such an influence has not reached the Semitic languages of the Arabian Peninsula.
Although reported discourse is overwhelmingly expressed directly, the construction is not
used for the expression of intention or goal. Thus, this is not an areal feature of the Red Sea
zone.
4.3 Agreement
As noted by Tosco (2000:349), Ferguson’s feature about the concord of plural nouns
with feminine singular in fact lumps together quite different phenomena: the more or less
regular agreement of plural nouns with adjectives, possessive pronouns and verbs in the
feminine singular in Semitic on the one hand, and the “gender polarity” in Cushitic in which a
noun inverts its gender in order to express the plural on the other hand. Looking at the
languages of our sample, the situation is even more diverse. Afar displays “gender polarity”
and number concord in the singular with plural nouns, plus a strict feminine singular concord
22
for names of peoples, but Beja has a strict concord rule for both gender and number, and no
gender polarity.
For Semitic languages the rule of feminine singular concord for plural nouns varies a
little from one language to another mainly according to the criteria of animacy. The Arabic
variety of Yafi‘ has a feminine singular concord for adjectives (not for verbs and pronouns)
with inanimate plural nouns, and a strict concord for both number and gender for animates.
But in Tigre the concord for plural inanimates is masculine singular34, and in Dahalik it is
also the case for animates. In the latter language, agreement between collective nouns and
verbs varies: one finds either feminine or masculine singular, or plural. For the other Semitic
languages, it is a matter of “free” variation. In MSA, concord in number, but not in gender,
may be omitted with deictics and relative pronouns. In the vernacular Arabic varieties of
north-east Africa, concord in both gender and number is unstable.
With numerals, concord rules in number with the counted noun vary. In Afar, Dahalik,
and Tigre the counted noun is singular (rarely plural in Tigre35), a common pattern
worlwide36, but in Beja it is always plural. The Cushitic SAM languages and Arabic have
number concord up to 10, but from 11 onwards, the counted noun is in the singular.
Such a variety means that we have to rule out Ferguson’s agreement feature for the
Red Sea zone.
4.4 Changes in the aspectual system: Auxiliaries
In all the languages, the aspectual systems based on a twofold distinction between a
perfect and an imperfect have been enriched by the introduction of periphrastic constructions
with auxiliary verbs. These auxiliaries, not surprisingly, are used for the expression of new
temporal, aspectual and modal values. In the course of history, some of these constructions
have become grammaticalized, like the suffix conjugation of Cushitic languages, and the
emergence of new verbal forms have reduced the old ones to dependent or modal uses.
Several semantic schemas have been used for the introduction of auxiliaries, but none
of them seem to be the result of a convergent tendency in the area as compared to other areas
or linguistic families. The auxiliaries which are productive today belong to the following
semantic categories: ‘to be’ (be, exist, become, stay) and ‘to have’, verbs of volition (want,
refuse), localization and movement verbs (put, go, come, stand, be seated, get up, stop, give),
‘to finish’, ‘to look for and not find’, ‘to say’, and ‘to do’. As far as we know, Beja uses by far
the highest number of auxiliary verbs in periphrastic constructions.
All languages have several verbs meaning ‘to be’, except Arabic which only has one,
and all these verbs are used in various kinds of verb phrases. They convey temporal and/or
aspectual values which vary according to languages and the morphology of the two
constituents: past in Arabic, MSA, Dahalik, Tigré, Afar and Saho, but not in Beja; progressive
in Beja, Tigre, Dahalik, Afar, but not in Arabic and MSA; also inchoative in Beja. Besides,
‘to be’ (in Beja) and ‘to have’ (in Afar) are also used in negative conjugations of some tenses.
Other auxiliaries often bring about values such as future, potential, negation, assertion,
or benefactive, and sometimes aspectual ones (progressive, imminence, durative, inchoative,
terminative).
The auxiliaries meaning ‘to say’ and ‘to put’, outlined in Ferguson’s article, have a
particular behavior because they do not only belong to the traditional TAM system: They are
34
Sometimes, if the inanimate noun is feminine, gender concord may occur.
Raz’ statement (1983: 13) appears to be somewhat contradicted by the texts he published, where plural
concords are quite numerous.
36
Tosco (2000: 347) adds: “… although actually rare in Semitic, and therefore of doubtful value as an areadiagnostic trait.”
35
23
a device for transcategorial and intracategorial derivation, used with different types of
morpho-syntactic categories: verbs, adverbs, nouns, ideophones and onomatopoeias. These
constructions are traditionally named descriptive compounds. For a complete and recent
analysis of the device, we refer to Cohen et al. (2002). It is perhaps interesting to note that
when there is a corresponding verbal base, the use of descriptive compounds falls within the
domain of discourse-participant interplay. Below there are a few examples which were not
included in Cohen et al. (2002):
Dahalik tubulluK bìla (splash he said) ‘he fell into the water’
Beja aan ee-kiley bess bir bir bir diin-aat-ka (these the-birds only pull pull pull say-VNDISTR) ‘These birds, each time they move their wings…’
As the construction is only found on the west bank of the Red Sea (in both AfroSemitic and Cushitic), and also exceed by far this geographical zone in Africa (Cohen et al.
2002: 227-251), it cannot be considered as an areal feature of the Red Sea zone.
This review shows that the syntactic features outlined by Ferguson and discussed by
Tosco are by no means more specific of the Red Sea area than of other areas in North-East
Africa or generally in Africa. What prevails on the whole is a great variety.
5
Conclusion
In spite of a very long common cultural and linguistic history, the above review has
shown that the Red Sea zone cannot be considered as a linguistic area, in the strict sense of
the term. Nevertheless it cannot be denied that deep influences and upheavals took place, but
in an unbalanced way. It seems that the Red Sea has been and still is a real borderline for
linguistic change in the Arabian Peninsula but not in Africa: major changes, such as language
shift, or more minor ones are all due to linguistic contact between the two banks of the Red
Sea; but they occurred in Africa, not in Arabia. Africa received a lot from Arabia, but the
reverse is not true. Thus the Red Sea zone is more accurately defined as a case of asymmetric
dominance relations between languages with large-scale shifts, to use the terms of Thomason
and Kaufman (1988: 95). But such a conclusion is only partial because it does not account for
what happened and is still happening on the west bank. As a matter of fact it seems that, on
the African bank, influences between Cushitic and Semitic have been and probably still are
bi-directional, and that all linguistic levels are concerned. In addition to the better known
substratum influence of Cushitic on Afro-Semitic, we underscored several times possible
Semitic influence on Cushitic for the retention of archaic traits such as the root-stem pattern
and ablaut, the position of the head in relative clauses, the absence of labio-velars. Still, in
this particular instance, mutual influences do not seem to reflect a long term multilateral
symmetric contact situation between the member languages of the west bank, a process which
is sometimes included in the characterization of Sprachbund. Rather they may mirror
successive stages due to the changing socio-linguistic situations and levels of bi- or multilingualism, and the prestigious role of Arabic because of religious, economic and political
reasons.
REFERENCES
Appleyard, David L. 1996. Ethiopian Semitic and South Arabian: towards a re-examination of
a relationship. in Izre'el, S. et Raz, S. (eds.), Israel Oriental Studies. Studies in Modern
Semitic languages, (vol XVI). Leiden. New York. Köln: Brill, p. 203-228.
Behnstedt, Peter. 1985. Die nordjemenitischen Dialekte. Teil 1: Atlas, (vol. 3). Wiesbaden:
Ludwig Reichert.
Behnstedt, Peter. 1987. Die Dialekte der Gegend von Sa'dah (Nord-Jemen), (vol. 1).
24
Wiesbaden: Harrassowitz.
Campbell, Lyle. 1994. Grammar: Typological and areal issues. in Asher, R. E. (ed.), The
Encyclopedia of Language and Linguistics, (vol 3.). Oxford: Pergamon Press, p. 14711474.
Cantineau, Jean. 1950. Racines et schèmes. in, Mélanges William Marçais. Paris:
Maisonneuve, p. 119-124.
Cohen, David. 1973-1979. "Qu'est-ce qu'une langue sémitique ?" Comptes rendus du
G.L.E.C.S., 1973-1979, vol. XVIII-XXIII, p. 431-461.
Cohen, David. 1984. La phrase nominale et l'évolution du système verbal en sémitique. Etude
de syntaxe historique. Leuven-Paris: Peeters.
Cohen, David. 1989. L'Aspect verbal. Paris: Presses Universitaires de France.
Cohen, David, Simeone-Senelle, Marie-Claude and Vanhove, Martine. 2002. The
Grammaticalization of "Say" and "Do": an Areal Phenomenon in the Horn of Africa. in
Güldemann, T. et von Roncador, M. (éds.), Reported Speech : A Meeting Ground for
Different Linguistic Domains. Amsterdam, Philadelphia: John Benjamins, p. 227-251.
Conti Rossini, Carlo. 1918. Schizzo del dialetto saho dell'alta Assaorta in Eritrea
(Nota).Estratto dai Rendiconti della R. Accademia dei Lincei, vol. XXII, fasc. 5º Seduta del 18 maggio 1913. Roma, Tip. della R. Accademia dei Lincei.
Crystal, David. 1987. The Cambridge Encyclopedia of Language. Cambridge: Cambridge
University Press.
Curnow, Timothy J. 2001. What Language features can be 'borrowed'? in Akhenvald, A. Y. et
Dixon, R. M. W. (eds.), Areal Diffusion and Genetic Inheritance. Problems in
Comparative Linguistics. Oxford, New York: Oxford University Press, p. 412-436.
Dahl, Östen. 2001. Principals of areal typology. in Haspelmath, M., et al. (eds.), Language
typology and language universals. Vol. 2. Berlin: de Gruyter, p. 1456-1470.
Dil, Anwar (ed.) 1980. Language and linguistic area. Stanford: Stanford University Press.
Dimmendaal, Gerrit J. 1992. Areal Linguistics. in Bright, W. (ed.), International
Encyclopedia of Linguistics, (vol. 1). Oxford: Oxford University Press.
Dimmendaal, Gerrit J. 2001. Areal diffusion versus genetic inheritance: an African
perspective. in Akhenvald, A. Y. et Dixon, R. M. W. (eds.), Areal diffusion and genetic
inheritance. Problems in comparative linguistics. Oxford, New York: Oxford
University Press, p. 358-393.
Doresse, Jean. 1983 (1971). Histoire sommaire de la Corne Orientale de l'Afrique. Paris:
Geuthner.
Dryer, Matthew S. 1989. Large linguistic areas and language sampling. Studies in Language,
1989, vol. 13-2, p. 257-292.
El Fasi, M. (ed.) 1997. Histoire générale de l'Afrique, III, L'Afrique du VIIe au XIe siècle.
Paris, Vanves: Présence Africaine, EDICEF, UNESCO.
Emeneau, Murray B. 1956. India as a Linguistic Area. Language 32, 1: 3-16.
Ferguson, Charles. 1976. The Ethiopian language area. in Bender, L., et al. (eds.), Language
in Ethiopia. Londres: Oxford University Press, p. 63-76.
Gilman, Charles. 1986. African areal characteristics: Sprachbund, not substrate? Journal of
Pidgin and Creole Languages, 1986, n° 1, vol. 1, p. 33-50.
Greenberg, Joseph. 1955a. Internal a-plurals in Afroasiatic (Hamito-Semitic). in Lukas (ed.),
Afrikanistische Studien Diedrich Westermann zum 80. Geburtstag gewidmet. Berlin:
Akademie Verlag, p. 198-204.
Greenberg, Joseph. 1955b. Studies in African Linguistic Classification. New Haven: Compass
Publishing Company.
Hagège, Claude et Haudricourt, André-Georges. 1978. La phonologie panchronique. Paris:
P.U.F.
Haudricourt, André-Georges. 1972. Problèmes de phonologie diachronique. Paris: SELAF.
25
Hayward, Richard J. 2002 (2000). Afroasiatic. in Heine, B. and Nurse, D., African
Languages. An Introduction. Cambridge, Cambridge University Press, p. 74-98.
Heine, Bernd. 1975. Language typology and convergence areas. Linguistics, 1975, vol. 144,
p. 27-46.
Heine, Bernd. 1976. A typology of African languages based on the order of meaningful
elements. Berlin: Dietrich Reimer.
Heine, Bernd. 1993. Auxiliaries. Cognitive Forces and Grammaticalization. Oxford: Oxford
University Press.
Heine, Bernd et Kuteva, Tania. 2001. Convergence and divergence in the development of
African languages. in Aikhenvald, A. Y. and Dixon, R. M. W. (eds.), Areal diffusion
and genetic inheritance. Problems in comparative linguistics. Oxford, New York:
Oxford University Press, p. 393-411.
Hetzron, Robert. 1977. The Gunnan-Gurage languages. Napoli: Istituto Orientale di Napoli.
Johnstone, Thomas Muir. 1977. Harsusi Lexicon and English-Harsusi Word-List. London:
Oxford University Press.
Johnstone, Thomas Muir. 1981. Jibbali Lexicon. London: Oxford University Press.
Johnstone, Thomas Muir. 1987. Mehri Lexicon and English-Mehri Word-List, with Index of
the English Definitions in the Jibbali Lexicon, compiled by G. Rex Smith. London:
SOAS.
Kurdi, Nafi H. 1994. L'Erythrée. Une identité retrouvée. Paris: Karthala.
Leslau, Wolf. 1938. Lexique soqotri (sudarabique moderne) avec comparaisons et
explications étymologiques. Paris: Klincksieck.
Leslau, Wolf. 1941. Documents tigrigna (Ethiopien septentrional). Paris: Klincksieck.
Leslau, Wolf. 1956. Etude descriptive et comparative du Gafat (Ethiopien méridional). Paris:
Klincksieck.
Leslau, Wolf. 2000. Introductory Grammar of Amharic. Wiesbaden: Harrassowitz.
Lonnet, Antoine and Simeone-Senelle, Marie-Claude. 1997. La phonologie des langues
sudarabiques modernes. in Kaye, A. (ed.), Phonologies of Asia and Africa, (vol 1).
Winona Lake, Indiana: Eisenbrauns, p. 337 - 372.
Masica, Colin P. 1976. Defining a linguistic area: South Asia. Chicago, London: University
of Chicago Press.
Miller, Catherine. 1991. De la cuisine familiale au commerce des spiritueux : Remarques sur
un parler de femme à Juba. in Kaye, A. S. (ed.), Semitic Studies in Honor of Wolf
Leslau ... (vol. 2). Wiesbaden: Harrassowitz, p. 1059-1084.
Morin, Didier. 1995. "Des paroles douces comme la soie". Introduction aux contes dans l'aire
couchitique (bedja, afar, saho, somali). Paris: Peeters.
Munro-Hay, Stuart. 2003. Arabia. in Uhlig, S. (ed.), Encyclopedia Aethiopica. Vol 1.
Wiesbaden: Harrassowitz, p. 294-300.
Parker, Enid M. and Hayward Richard J. 1985. An Afar - English-French Dictionary (with
grammatical notes in English). University of London, School of oriental and African
Studies.
Perrot, Jean and Cohen, David (eds.). 1988. Les langues dans le monde ancien et moderne.
Langues chamito-sémitiques. Paris: Editions du CNRS.
Pliny the Elder. Natural History, 10 vols. London: Harward University Press, 1992. Latin text with English
translation.
Primus, Beatrice. 2001. Word order typology. in Haspelmath, M., et al. (eds.), Language
typology and language universals. Vol. 2. Berlin: de Gruyter, p. 855-873.
Prouty, Chris and Rosenfeld, Eugene. 1994. Historical Dictionary of Etiopia and Eritrea.
London: The Scarecrow Press.
Raz, Shlomo. 1983. Tigré Grammar and Texts, (vol 4). Malibu: Undena Publications.
26
Reinisch, Leo. 1878. Die Sprache der Irob-Saho in Abessinien. Wien, Karl Gerold's Sohn.
Reinisch, Leo. 1893. Die Bedauye-Sprache in Nordost-Afrika. Wien: Sitzungs berichte der
Kais. Akademie der Wissenschaften in Wien L, Band CXXVIII.
Robin, Christian (ed.) 1992. L'Arabie antique de Karib'îl à Mahomet. Nouvelles données sur
l'histoire des Arabes grâce aux inscriptions. Revue du Monde Musulman et de la
Méditerranée, (vol 61). Aix-en-Provence: Edisud.
Roper, E.M. 1928. "Tu Bedawie". An Elementary Handbook for the Use of Sudan
Government Officials. London: Austin.
Ross, Malcolm D. 1996. Contact-induced change and the comparative method: Cases from
Papua New Guinea. in Durie, M. (ed.), The comparative method reviewed. Regularity
and irregularity in language change. New York, Oxford: Oxford University Press, p.
180-217.
Sasse, Hans-Hürgen. 1986. A southwest Etiopian language area and its cultural background.
in Fishman, J. A., et al. (eds.), The Fergusonian Impact: In Honor of Charles A.
Ferguson on the Occasion of His 65th Birthday. Berlin: Mouton de Gruyter, p. 327-342.
Schoff, Wilfred H. (ed.) 1912. The Periplus of the Erythraean sea. Travel and trade in the
Indian Ocean by a merchant of the first century. New York, London, Bombay and
Calcutta: Longmans, Green ,and Co.
Simeone-Senelle, Marie-Claude. 1996. Le fonctionnement des verbes de mouvement dans des
dialectes de la Tihama du Yémen. in HOLES, C. (ed.), 2nd AIDA Conference. 10-14
sept. 1995. Trinity Hall. University of Cambridge. Angleterre: Faculty of Oriental
Studies. University of Cambridge, p. 227-235.
Simeone-Senelle, Marie-Claude. 1997. The Modern South Arabian Languages. in Hetzron, R.
(ed.), The Semitic Languages. London: Routledge, p. 379-423.
Simeone-Senelle, Marie-Claude. 1998. La dérivation verbale dans les langues sudarabiques
modernes. Journal of Semitic Studies, 1998, vol. xliii/1, p. 71-88.
Simeone-Senelle, Marie-Claude. 2000. La situation linguistique dans le sud de l'Erythrée. in
Wolff, E. H. et Gensler, O. D. (eds.), Proceedings of the 2nd World congress of African
Linguistics, Leipzig 1997. Cologne: Rödiger Köppe, p. 261-276.
Simeone-Senelle, Marie-Claude. 2000. Les Langues en Erythrée. Chroniques Yéménites,
2000, p. 167-173.
Simeone-Senelle, Marie-Claude. 2000. L'arabe véhiculaire parlé en Erythrée sur la côte de la
mer Rouge, de Massawa à Rahayta. in Bettini, L. (ed.), Studi di diallettologia Araba
(Oriente Moderno, n.s. XIX (LXXX) 1). Roma: C.A. Nallino, p. 153-181.
Simeone-Senelle, Marie-Claude. 2002. Les langues sudarabiques modernes à l'aube de l'an
2000: Evaluation des connaissances. in Izre'el, S. (ed.), Semitic Linguistics: The state of
the art at the turn of the 21st Century. Israel Oriental Studies XX. Tel Aviv:
Eisenbrauns, p. 379-400.
Simeone-Senelle, Marie-Claude. 2002. L'arabe, langue maternelle de citoyens djiboutiens du
nord de la République de Djibouti. in Youssi, A., et al. (eds.), Aspects of the dialects of
Arabic today. Proceedings of the 4th Conference of the International Arab Dialectology
Association (AIDA), Marrakesh, Apr. 1-4. 2000. In Honour of Professor David Cohen.
Rabat: AMAPATRIL, p. 140-150.
Simeone-Senelle, Marie-Claude. to appear. Dahalik Language. in Uhlig, S. (ed.),
Encyclopaedia Aethiopica, (vol 2). Wiesbaden: Harrassowitz.
Simeone-Senelle, Marie-Claude. in press. Les langues sudarabiques modernes. in Bonvini, E.
et Peyraube, A. (eds.), Encyclopédie des sciences du langage. Dictionnaire des langues.
Paris: PUF.
Simeone-Senelle, Marie-Claude. to appear. Some characteristics of Dahalik, a newly
discovered Afro-semitic language spoken in Eritrea. in Uhlig, S. (ed.), Proceedings of
the15th International Conference of Ethiopian Studies. Hamburg, July 2003.
Simeone-Senelle, Marie-Claude and Vanhove, Martine. 2003. Le fonctionnement d'auxiliaires
en afar. in Lentin, J. et Lonnet, A. (eds.), Mélanges David Cohen. Etudes sur le
27
langage, les langues, les dialectes, les littératures, offertes par ses élèves, ses collègues,
ses amis. Présentés à l'occasion de son quatre-vingtième anniversaire. Paris:
Maisonneuve et Larose, p. 615-34.
Thomason, Sarah G. 1986. Contact-induced language change: possibilities and probabilities.
in Boretzky, N., Enninger, W. and Stolz, T. (eds.), Beiträge zum 2. Essener Kolloquium
über "Kreolsprachen und Sprachkontakte". Bochum: Brockmeyer, p. 261-284.
Thomason, Sarah Grey. 2001. Contact-induced typological change. in Haspelmath, M., et al.
(eds.), Language typology and language universals. Vol. 2. Berlin: de Gruyter, p. 16401648.
Thomason, Sarah Grey and Kaufmann, Terrence. 1988. Language contact, creolization, and
genetic linguistics. Berkeley, Los Angeles, London: University of California Press.
Tosco, Mauro. 2000. Is there an 'Ethiopian language area'? Anthropological Linguistics, 42,3,
p. 329-365.
Tuscherer, Michel (ed.) 1993. Le Yémen, passé et présent de l'unité (Revue du Monde
Musulman et de la Méditerranée, 67/1). Aix-en-Provence: Edisud.
Vanhove, Martine. 1995. Notes on the Arabic Dialectal Area of Yafi'.(Yemen). Proceedings
of the Seminar for Arabian Studies (Oxford, July 1994), 1995, vol. 25, p. 141-152.
Vanhove, Martine. 1997. Diversité dialectale et nivellement linguistique : Le cas de l'arabe au
Yémen (sud et est). in Caron, B. (éd.), Proceedings of the 16th International Congress
of Linguists, Paris 20-25 July 1997. Paris: Pergamon, Oxford, p. Paper n. 115.
Vanhove, Martine. 2002. Processus de koïnisation dans le dialecte arabe de Yafi‘ (Yémen). in
Youssi, A., et al. (éds.), Aspects of the Dialects of Arabic Today. Proceedings of the 4th
Conference of the International Arabic Dialectology Association (AÏDA) Marrakesh
Apr. 1-4 . 2000). In Honour of Professor David Cohen. Rabat: Amapatril, p. 151-159.
Vanhove, Martine. 2003. Bilinguisme et alternance bedja-arabe au Soudan. in Ferrando, I. et
Sánchez Sandoval, J. J. (eds.), AIDA 5th Conference Proceedings. Cádiz, September
2002. Cádiz: Servisio de Publicaciones Universidad de Cádiz, p. 131-142.
Vanhove, Martine. to appear. Arabic of Yemen. in Versteegh, K. et Köndgen, O. (eds.),
Encyclopedia of Arabic Language and Linguistics (EALL). Leiden: Brill.
Vanhove, Martine and Hamid Ahmed, Mohamed-Tahir. in press. Le bedja. in Bonvini, E. et
Peyraube, A. (eds.), Encyclopédie des sciences du langage. Dictionnaire des langues.
Paris: PUF.
Vergari, Moreno et Vergari, Roberta. 2003. A Basic Saho-English-Italian Dictionary. With an
English-Saho Index and Grammatical Notes (by Giogio Banti and Moreno Vergari).
Asmara: The Authors.
Zaborski, Andrzej. 1975. The Verb in Cushitic. Krakow: Nakladem Uniwersytetu
Jagiellonskiego.
Zaborski, Andrzej. 1986. The Morphology of Nominal Plural in the Cushitic Languages, (vol
28). Wien: Beiträge zur Afrikanistik.
Zaborski, Andrzej. 1991. Ethiopian Language Subareas. in Pilaszewicz, S. et Rzewuski, E.
(eds.), Unwritten Testimonies of the African Past. Warsaw: Wydawnictwa Uniwersytetu
Warszawskiego, p. 123-134.
ABBREVIATIONS
A
accusative
DIST
distributive
O
pf.
pl.
f.
feminine
G
genitive
H
head noun
ipf.
imperfect
m.
masculine
MSA Modern South Arabian
POST
S
sg.
subj.
V
28
object
perfect
plural
postposition
subject
singular
subjunctive
verb
VN
verbo-nominal
29