Publications

The Southern Cushitic languages in Tanzania have in their sound inventories two lateral obstruents: a voiceless lateral fricative and an ejective lateral affricate. Both sounds can be shown to be inherited from Cushitic and Afroasiatic and are reconstructed with lateral fricative pronunciation. Yet the Southern Cushitic languages are the only Cushitic languages that maintain the lateral fricative pronunciation. In the area where these languages are and were spoken, there are a number of other unrelated languages with these lateral sounds: Hadza (isolate) has a voiceless and voiced lateral affricate and a lateral fricative, Sandawe (unclassified) likewise has a lateral fricative and a lateral affricate, and Southern Nilotic has been reconstructed with a lateral fricative that is only documented in a now extinct primary branch. The feature has been proposed as one diagnostic of the Tanzanian Rift Valley as a linguistic area. In such a scenario, the presence of laterals in the area could be interpreted as contact-supported retention of otherwise versatile sounds. Additional language groups in East Africa in which laterals are attested are Kuliak (Nilo-Saharan or unclassified) in Uganda, consisting of three languages, and Taita Bantu languages in Kenya. In their case, the challenge is to determine the origin of the sounds (inheritance, contact, or innovation). This paper investigates the historical processes within language contact that resulted in the areal distribution of lateral obstruents in East Africa.

Keywords: lateral obstruents, contact-supported retention, areal linguistics, Tanzanian Rift Valley language area

This chapter examines the imact of contact between Rangi-speaking and Alagwa-speaking communities. Rangi is a Bantu language spoken by some 310,000 people in central Tanzania. Alagwa is a Cushitic language spoken by approximately 52,000 people in the same region. Histories of central Tanzania suggest that these languages have long been in contact, with this situation continuing to the present day. We report here on the shared history and interactions between the groups, focusing particularly on the social domains of family, kinship and marriage where there is evidence of sustained contact. We explore also the broader community context and social exchange, as well as acknowledging the role of Swahili as the language of wider communication throughout the region.

Kiessling and Mous (2023:38-40) show specific influence of an early form of Oromo on proto-West-Rift-South-Cushitic comprising 14 lexical items and 2 grammatical morphemes. In the meanwhile, we have found additional evidence substantiating this claim. Given the geographical position of Oromo and its expansion history, such a contact can only be imagined in Kenya and less than 1000 years ago.

We present our intermediate results in the project Linguistic History of East Africa, http://www.lheaf.org The school book version of East Africa’s history is that the first inhabitants were Hunter-Gatherers with a stone-age culture speaking click languages surviving in Hadza and Sandawe. Cushitic people moved from Ethiopia into East Africa long ago bringing animal husbandry and agriculture; 4000 years ago; linked to the Pastoral Neolithic archaeological culture. Later came Bantu people from the West, the area of the great lakes, bringing more agriculture and iron, 1500 years ago. Still later, from the north, South Sudan, arrived the Nilotic speaking people with their cattle culture, 1000 years ago.

A crucial consideration is the debate whether Tanzanian Cushitic is a primary branch of Cushitic or a sub-sub branch of East Cushitic. And this is the issue that we address in the presentation. We show the presence of specific pre-Oromo lexical elements in Tanzanian Cushitic. We show the shared innovation of plural markers in Tanzanian Cushitic and East Cushitic. We discuss the challenges and potential consequences of classifying Tanzanian Cushitic as Oromoid and we propose a complex history of Tanzanian Cushitic involving more than one Cushitic stratum. Finally, we show how these issues relate to the various other historical issues that we address in our project.

Various cattle terms in East African Bantu languages are Cushitic in origin. Prominent
examples are ndama ‘calf’ and maziwa ‘milk’ in Swahili and in many other Bantu languages
of East Africa. We intend to trace Cushitic cattle terminology in Bantu languages: to
determine when and where these words entered the predecessors of the current Bantu
language groups, but also in order to trace the further spread from Bantu language to another
by checking which later developments were regular and irregular (sign for cross Bantu
borrowing). In this we ultimately seek to trace the Bantu expansion within East Africa.
Moreover, we hope to recover cultural change. In this article we present the first results on
distinctive terms for male domestic animals. We show that East African Bantu languages
innovated terms for male domestic animals and an important source was from South Cushitic
partly before this group of languages entered Tanzania suggesting East of Lake Nyanza as
area of contact where the arriving Bantu groups met settled Cushitic communities. We
suggest that the innovation of words for precisely male domestic animals points to Bantu
innovation on breeding as cultural practice under Cushitic influence.

The Tanzanian Cushitic languages have a word for ‘three’ *tam ‘3’ that is cognate with the word for ‘two’ *lam ‘2’ in the East Cushitic branches (Sasse 1979); and a word for ‘four’ that is cognate to the word for ‘three’ in East Cushitic, while the terms for ‘five’ and ‘six’ correspond regularly in form and meaning with the rest of Cushitic. This remarkable semantic shift in itself shows that the East-Rift SC language Aasa which has the regular reflex for ‘two’ did not share this intriguing innovation, and is therefore not part of West-Rift; the other East-Rift SC language, Kw’adza, does share this 2 > 3 semantic innovation, and is better classified with the rest of Tanzanian Cushitic. A secondary interesting historical number story is that of haka ‘4’: a number of unrelated languages in the region have this root for ‘4’: Aasa and Qwadza (both South Cushitic, Tanzania) have hak. The mixed register of the Bantu language Inner Mbugu (or Mixed Ma’á) has hai. Sandawe has haká ‘4’ which is a reflex of Khoekhoe *haka ‘4’ (Voßen 1997:503). This points to a Sandawe transfer into Qwadza, Aasa, and eventually Mixed Ma’á. These are all languages that are not directly related nor in direct contact presently. However, the word is also used in Gorwaa and Iraqw (Tanzanian Cushitic) in the register of diviners counting stones; showing that is more widely known in the area. This requires scenarios of earlier contact. The historical semantic shift in the numbers of West-Rift South Cushitic suggest a history that involves more than one Cushitic branch, and more than one language migration. The wide distribution of haka shows the earlier importance of Sandawe as a donor language with prestige.

In his seminal work on the classification of African languages, Greenberg (1963) has South Cushitic as one of the primary branches of Cushitic. This proposal has been immensely influential in the interpretation of the history of East Africa. On the basis of this classification, the assumption is that that the presence of South Cushitic in Tanzania is ancient (Ehret 1980), and pre-dates the entry of Bantu and Nilotic peoples (Ambrose 1998). As a consequence, (South) Cushitic has been linked to the Savannah Pastoral Neolithic cultural complex that is recognised in archaeology, and it has been proposed that the introduction of agriculture and cattle-keeping in Tanzania can be attributed to the South Cushitic speakers. Ehret’s (1980) reconstruction of South Cushitic lexicon and phonology has been extremely influential in recognising linguistic contact in many of East Africa’s Bantu and Nilotic languages despite the fact that this reconstruction has been severely criticised (Philippson 2013). Serious doubts on the classification of South Cushitic as primary branch have been raised though. Hetzron (1980) has pointed to many grammatical resemblances between South Cushitic and East Cushitic languages that argue for inclusion of South Cushitic within East Cushitic. In Tosco’s (2000) Cushitic overview this uncertainty is represented by the fact that South Cushitic figures differently in his genetic trees: as primary branch of Cushitic for the classical view and as primary branch within East Cushitic reflecting Hetzron’s suggestions; the issues are discussed in detail in Kießling (2001). Kießling and Mous (2003) provide an extensive lexical and phonological reconstruction of the four South Cushitic Tanzanian languages that are still spoken; while Kießling (2002) is a detailed morphological reconstruction of these languages. Kießling and Mous (2003) pointed out wider Cushitic parallels where they could but this did not enable them to suggest a position of Tanzanian Cushitic in the Cushitic tree. The challenges are the lexical innovations that Tanzanian Cushitic must have undergone and for many of these no other languages could be suggested as sources. In addition, the other languages that were classified as South Cushitic do not offer much for reconstruction: the languages Aasáx and Qwadza are obsolete and the data on these are unreliable because they were collected from rememberers rather than speakers (Kruijsdijk 2023); Ma’á, often characterised as a mixed language, is Bantu, and not Cushitic, and some of the original Cushitic lexicon after language shift survives in a parallel register which also contains words from a variety of other sources including manipulated words from the basis Bantu vocabulary (Mous 2003). The last suggested member, Dahalo, is now considered to be (marginally) East Cushitic rather than South Cushitic (Tosco 1989, Tosco and Blazek 1994).

                I propose that Tanzanian Cushitic is a primary branch of Cushitic after all. I also argue that the earliest South Cushitic expansion into Tanzania was followed by two others that have left their (lexical) impact on Tanzanian Cushitic. The latest is the pre-Oromo influence on Tanzanian Cushitic. The suggestion for such language contact showing transfer from pre-Oromo lexical and morphological material featured already in Kießling and Mous (2003), and was recently substantiated by Rapold (2023). There is plenty of reconstructed Tanzanian Cushitic material that is clearly Cushitic but did not undergo the Oromoid innovations. An earlier expansion is formed by speakers of the Dullay-Yaaku subgroup (see Hayward 1978 that this is a subgroup). Recently Sands and Tosco (2022) have argued that early Dullay-Yaaku speakers must have been in contact with Hadza (a language isolate and in the area of Tanzanian Cushitic). I provide further evidence for this intrusion by showing Dullay-Yaaku influence on proto-Tanzanian Cushitic while the Tanzanian Cushitic proto lexicon also contains Cushitic lexical evidence that pre-dates Dullay-Yaaku. The consequences for the interpretation of East Africa’s history are far-reaching: There was not one migration of Cushitic speakers into Tanzania but at least three. For all Cushitic lexical transfer into Bantu and Nilotic languages of Tanzania and Kenya, the source needs to be considered.

I present the main results of the LHEAf project, www.lheaf.org on this topic arguing 1. South Cushitic is not part of East Cushitic, 2. Proto Tanzanian Cushitic is influenced by Oromo in Kenya, 3. Taita Cushitic is South Cushitic, 4. (Tanzanian) South Cushitic and South Nilotic / Datooga were in contact. On the basis of the linguistic history I discuss possible places and history of people.

Ehret (1971:114-118) in his PhD thesis on the history of Southern Nilotic pointed to South Cushitic loans in proto-Southern-Nilotic and proto-Kalenjin. In view of later lexical reconstructions (i.a., Rottland 1984, Kiessling and Mous 2003), lexical description of the relevant languages, we are able to confirm and strengthen this proposal and expand the number of loans. This shows profound interaction between proto-Kalenjin and Pre-Proto-West-Rift-South-Cushitic. Such contact is likely to have taken place in West-Kenya.

In a team representing experience in various relevant languages and language families (Ed Elderkin, Maarten Mous, Derek Nurse, Christian Rapold, Bonny Sands, Ahmed Sosal, Mauro Tosco, Sjef van Lier), we meet every other Monday for two hours to find the origin of the sizeable number of borrowed items in the Bantu languages of the Taita mountains and related areas. After a critical re-assessment of the earlier proposal (Ehret and Nurse 1981), building on Sophie Mulder’s BA thesis, we are now working through vocabulary that seems not inherited from proto-Bantu.