The limited evidence of shared innovations in East Cushitic

The traditional approach to language classification has long relied on evidence from shared innovations, to the extent that many may not realise how problematic such evidence can be. After first reviewing the issue in general, the present paper point out the limitations of evidence for shared innovations in East Cushitic languages, both in general terms and in the particular case of heavy diffusion between Dullay and Konsoid varieties. It concludes by stressing the need to use all evidence available for subgrouping, both lexicostatistical and any available on shared innovations.