The 27 or so lects grouped into the SBW sub-group of Chadic languages (Shimizu, 1978) are spoken in the South of the Bauchi state (Nigeria) stretching from Bauchi to the south of the state over less than 100 kms. Among them, only Zaar is relatively less unknown (Caron, 2002; Schneeberg, 1971, 1974) and is often used to characterise the whole SBW group. These languages or dialects are obviously closely related when their basic vocabulary is compared. However, various elements tend to set Zaar (Saya) and Gùùs (Sigidi) apart : * From a socio-linguistic and historical point of view : the name ßarawa used by Shimizu to name the group is not acknowledged by the Zaar (Sayawa) and Gùùs (Sigidi) people. If all the ßarawa agree to be closely related and all originate from the same area, viz. the Buli and Zaranda hills near Bauchi, the Zaars say they came from Bornu through Zaria ; they say they settled in the hills South of Tafawa Balewa after some cohabitation with the Angas, Pyem and Jarawa. * Zaar and Gùùs have three tone levels, the others only two. * Zaar and Gùùs have a very elaborate and profusive TAM system when compared to the other languages. All these elements single out the SBW group of languages for what Petr Zima has called a laboratory for the study of language evolution. This gives us an opportunity to first describe the extraordinary variation among languages that are so small in terms of number of speakers, and so close in terms of geographical and genetic proximity, and then to try and account for this variation. In order to do so, we have picked one language from the Geji, Polci and Zeem clusters (resp. Bùù, Nyámzàk and Cààrí) and two from the Dass and Saya clusters : Baraza and Zòɗì (Dott) for Dass ; Gùùs (Sigidi) and Zaar for Saya.