Introduction

It has been recognised for many decades that Boro, spoken in the plains of northwestern Assam, is a tone language, but there has been a conspicuous lack of agreement about just how many tones need to be recognised. Halvorsrud (1959) and Burling (1959) gave the language only two tones. Burton-Page (1955) described it as having three tones, while Bhattacharya (1977) gave it no less than four. In a recent monograph (2006), in which we compared the phonological systems of Boro, Garo, Tiwa, and Rabha, we implicitly confirmed the difficulty of the Boro tone system. We felt that we had a good understanding of the tones of Tiwa and Rabha, and of the cognate glottal stops of Garo, but the tones of Boro proved to be more difficult. The tones of one syllable Boro words seemed clear, but the tones of longer words proved to be very difficult, and, we were forced, in our monograph, to omit tone marks from Boro polysyllables. Something was special and different about Boro tones that prevented us from arriving at a satisfactory analysis.

In the winter of 2006, we had the opportunity to return to Boro and we can report considerable progress. In this article, we describe the tones of polysyllabic Boro words as we now understand them. Our new understanding also throws some light on why different observers have come to such different conclusions about the number of tones in the language.