The Sanskrit dhātu brū ‘to say, speak, tell’ (ṚV+) is usually connected to the Iranic root mrao– ‘to say, speak’ (Av). Both the standard etymological dictionary of Sanskrit (EWAia-II: 235-236) and the handbooks of Indo-European (e.g. LIV2: 446-447) derive it from the Proto-Indo-European root *mleuh2– ‘to speak’.
Nevertheless, this etymology is problematic due to the passage *ml– > *mr– > *br-, which is usually explained by arbitrary assumptions (e.g. “br° beruht wahrscheinlich auf den Sonderbedingungen eines Verbums für ‘sagen’”, EWAia-II: 236; followed by LIV2: 446).
The aim of this paper is to show how the passage *mr– > *br– is irregular in Sanskrit, on the basis also of Tocharian and Slavic data. This observation points to the possibility that in fact we are dealing with two different verbs: the root *mleuH– continued in Avestan and possibly in Slavic, and a Proto-Indo-European root *bleuh2– ‘to say, speak, tell’ continued both in Sanskrit brū (Vedic brav-) and Tocharian pälw-.