Wednesday, January 9, 2008

Nilo-Saharan and the Sahel

So, it seems as though (from my relatively amateur perspective) languages of the Sahara and Sahel are particularly hard to classify, whether because of the relative lack of data already mined on them, or because of the inherent difficulties in the languages in finding conclusive similarities. Sociologically and anthropologically speaking, both of these probably have a lot to do with the relatively sparse populations in the region. The more contact, the more linguistic similarity one would presume, and the less contact, the less discernible links - except the region has historically been peopled by nomads and traders who traverse vast expanses and therefore have contact with languages from the Mediterranean to the jungle, and the Atlantic to the Indian Ocean. The dramatic climatic changes over time also complicate things to some extent, especially the further back you go. So the classification of many languages, or clusters of languages, is very unresolved.
The classification of languages could seem esoteric and irrelevant, but what if it could lead to a decrease in ethnic tensions, through more widespread understanding of the common roots of groups that see themselves as separated and opposed since time immemorial? This particularly tends to be the case between nomads and more sedentary groups, who are often so culturally tied to their occupational identities that they associate with any group of that occupation more than neighboring ethnic groups with whom they have lived in symbiosis for years. My favorite example of this sort of is a Fulani (Fuuta Toro) friend who said that the shepherds to whom the birth of Jesus was first announced by angels (as in the famous christmas carol “Angels we have Heard on High”), were Fulani. He stopped short of saying that King Herod (who tries to kill baby Jesus later in the story) was a Beydhani, but the recent memory of les évènements (’87 – ’91) had definitely colored his memory of centuries of co-existence as bitter.

On the other side of the Sahel, here in Darfur, we are seeing the extermination and absorption of minority ethnic groups, and what has been termed ethnic cleansing of non-Arabs. In Alexander De Waal’s book “Darfur: A short history of a long war,” (which is by far the best and most in-depth analysis of the history of a convoluted and confusing conflict) he reveals internal documents of the nomadic Arabic-speaking tribes, which indicate that they see themselves as Sudan’s only pure Arabs, and that their attempts to purify Darfur of non-Arabs is only the first step toward purging the entire country’s power structure of the dominant riverain Arabs.

Perhaps I am too hopeful in the possibility of linguistic analysis bringing about change, but this is the motivation that pushes me to explore the relationships of some of these languages, and of particularly are the Nubian languages, Meroitic, and Fur. Perhaps a later analysis will look at wider Sahelian trends through language groups such as Fulani, Tuareg, and Songhay (maybe with some help from Jabal al-Lughat and the Ideophone?). Specifically, my question is how the Nubian languages could be so widespread geographically within Sudan without evidence of other attendant elements of the culture having expanded, and without having had a greater overlap with Meroitic.

The common historical reconstruction of what happened is that at some point long ago (at le, the Nubian languages dispersed, and a major part of the family ended up in the Nile Valley and succeeded what remained of the Meroitic Empire. This is covered in the proceedings of a 1981 conference organized by Ehret and Posnansky, the reviewer of whose publication articulates thus:
“When and how [was] the Meroitic kingdom on the Nile penetrated and largely resettled by Nubians from Kordofan and Darfur? The conclusion reached is that, rather than a process of rough and rapid conquest narrowly anticipating that by the Aksumites of King Ezana, most of the Nubians arrived as peaceful settlers five or six hundred years earlier, at a time when the country between the first and third cataracts was temporarily underpopulated. The idea which is persuasively presented by William Adams, seems to have occurred to him in connection with the absence of archeological data for the first millennium BC revealed by all the elaborate surveys undertaken in preparation for the Aswan High Dam. It took shape when he studied the glottochronological evidence for the divergence of the Nobiin and Dongolawi languages, for which a linguistic paper by Robin Thelwall provides the wider background.”

Yet this does not fit with the obvious textual evidence that shows a gap between Meroitic and Nubian cultures (Note that there is evidence of Nubian inscriptions before the 8th C. but not “texts” per se):

“The corpus of Meroitic Late texts can be dated to circa 5th century CE whereas the Old Nubian corpus extends from the 8th century CE (Browne 2002). The texts of Old Nubian and Meroitic are only distanced by a few hundred years. Diachronically this small length of time would allow us to see the relatedness of these languages if indeed they were, although the attempts to position Meroitic as an ancestor of Old Nubian have always resulted in disappointment for those who have chosen to pursue this line of investigation.” (Rowan 2006)

Though we can determine that Meroitic and Nubian are not very related linguistically, it seems likely that Nubia appropriated a lot of the imperial infrastructure left over from Meroe. They almost certainly did not co-exist along the Nile for the 1500 years that Adams proposes before the Nubians suddenly rose to prominence. However it is also unlikely that the Nubian languages spread to their current dispersal before they developed their writing system, given Arkell’s findings of Nubian inscriptions and pottery shards as far west as Dongola and Eastern Darfur, which one would have to date to (at earliest) the 10th c. CE. This would seem to indicate a late dispersal date, but what drives the assumption that all the Nubian population movements occurred simultaneously? The Nuba Mountains branch of Nubian languages including Dilling, Ghulfan, and Kadaru seems likely to have been the center of gravity of the family, a theory supported by the fact that there exist the most number of dialects of Nubian here in a relatively small geographical area (the Nuba Mountains seem to be an “accretion zone” as Nichols describes it), and that there are no traces of any writing or other material cultural influence in these areas.

“For a long time it was assumed that the Nubian peoples dispersed from the Nile Valley to the south, probably at the time of the downfall of the Christian kingdoms. However, comparative lexicostatistic research in the second half of the twentieth century has shown that the spread must have been in the opposite direction (Thelwall 1982, Adams 1982, among others). Greenberg (as cited in Thelwall 1982) calculated that a split between Hill Nubian and the Nile-Nubian languages occurred at least 2,500 years ago. This account is corroborated by non-linguistic evidence — for example, the oral tradition of the Shaiqiya tribe of the Jaali group of arabized Nile-Nubians tells of coming from the southwest long ago.” (Wiki)

In fact the Shaiqiya are quite significant further for the linguistic evidence found among a subgroup, the Nidayfab, which preserved many Nubianisms in a collection of legal documents from the early 19th c. See Jay Spaulding’s “The Old Shaiqi Language in Historical Perspective.” for more of the specifics, but the key point is that until relatively recently, the Shaiqi and all riverain Arabs saw their identity as autochthonously Sudanese (and thus part of the Meroitic- Nubian Political continuity) rather than Arab.


So perhaps we can say that about 2500 years ago, a large number of Nubians went from the Nuba Mountains northward, but did not arrive to the area which the Meroitic empire was inhabiting until after it was destroyed by Aksum, at which point they saw an opening to establish themselves in the vacuum left behind… but in order to fill the large shoes left behind for them, they had to have a language and material culture to match their predecessors, but wanted to be a part of the global economy, which required Greek, so they borrowed Greek letters via Coptic and added 3 Meroitic letters for good measure, putting forth a standard literary language. This clearly would not have reflected the diversity of the different dialects which composed the greater Nubian empire (actually 3 sub-empires, of Nobatia, Makuria, and Alodia), and the limited nature of Nubian writings (religious texts, political proclamations, and a few contracts) makes it likely that their was a vast discrepancy between spoken and written Nubian, and that this diglossia reflects the current diversity in modern riverain Nubian dialects, namely between Kenuzi-Dongolawi, and Nobiin, which probably represent two different empires. This diglossia and diversity of dialects also may explain why the subsequent move of the mysteriously stranded Darfurian dialects of Nubian could have occurred so late, and been accompanied by material evidence (the above mentioned inscriptions detailed by Arkell) and seem so lexically distant from the other strands. I suggest that not only did Midob and Birgid (now extinct) leave toward the waning of the Nubian empires in response to pressure from without, but that these were not the only Nubian languages to have moved out there, and that others have become extinct (like Birgid), and others remain to be classified as Nubian languages… And here’s where it gets crazy – I think the Fur language has significant Nubian influence both lexically and otherwise. The (as-yet unpublished) Fur dictionary I have is poorly organized and not conducive to searching, so more in depth analysis will be forthcoming from that angle, but superficially speaking, Fur and the Nubian languages both have a limited range of tones, they share some plural formations (Fur has multiple) and the ascendancy of the Fur Sultanate not long after the Nubian empire’s demise, in the midst of a jumble of different ethnic groups smacks of some sort of transmission of imperial memory.


To take it back to the meta-level and summarize, a language family (Nubian) within the Nilo-Saharan phylum originates in an accretion zone (Nuba Mountains) with contact with a number of Niger-Kordofanian* language families, then moves into an area dominated by languages within the Afro-Asiatic stock* (Meroitic, then Arabic) until most of the speakers adopted the dominant Semitic language. Some however migrated to an area which was again a Niger-Kordofanian dominated accretion zone (Jebel Marra) where their languages became heavily influenced by these languages until the dominance of Arabic again overtook them. The repeated and prolonged contact of this Nilo-Saharan language family with Niger Kordofanian languages contributes to the extensive lexical similarities noted by Blench in his designation of a Niger-Saharan macrophylum.

*Johanna Nichols has much more on the classification of "quasi-stocks" Niger-Kordofanian and Afro-Asiatic here:

The type-defining example is probably Niger-Kordofanian, a set of families
and stocks mostly of sub-Saharan Africa including the Bantu family (Bendor-
Samuel 1989, Greenberg 1963), discussed below. The genetic marker of
Niger-Kordofanian is its complex systems of generally prefixal genders (also
called concord classes), in which there are particular prefixes for particular
classes and systematic correspondences between singular and plural concord
classes. The system is shared widely among the daughter branches and is identifiable
as a system even when individual elements are greatly changed or lost.
This kind of gender system is quite specific and quite rare worldwide and thus
useful as a genetic marker. Yet systematic sound correspondences and regular
lexical reconstructability are absent from Niger-Kordofanian, the internal
structure of the genetic tree is still in doubt...
An atypically stock-like quasi-stock is Afroasiatic, another African group
established by Greenberg (1963). It consists of the families Chadic and Berber,
the isolate Egyptian, the Semitic stock, a likely isolate Beja, a stock or pair of
stocks Cushitic, and possibly the family Omotic (see e.g. Bender 1989, Newman
1980). The Afroasiatic quasi-stock has a distinctive grammatical signature
that includes several morphological features at least two of which independently
suffice statistically to show genetic relatedness beyond any reasonable
doubt (the entire set is listed inNewman 1980; for statistical significance,
see Newman 1980 and Nichols 1996). Hence it is routinely accepted as a genetic
grouping, though uncontroversial regular correspondences cannot be
found and a received reconstruction may never be possible.

1 comment:

David Marjanović said...

Nichols' pessimism ("cannot be found" instead of "haven't been found") is strange.