Although applied to the description of languages as diverse as Ewe of Ghana (Clements 1978), Mixtec of Mexico (Hunter and Pike 1969), and Kairi of Papua New Guinea (Newman-Petterson 1990), the term tone sandhi refers, first and foremost, to the kind of tonal alternations one typically finds in languages spoken in China and surrounding areas of the southeast Asian mainland. These languages share certain areal characteristics, notably highly developed tonal repertoires matched, in many cases, by even more complex tonal alternations in connected speech. The Sanskrit root of sandhi means junction, connection, combination, or liaison (cf. Allen 1962, Andersen 1986). Tone sandhi, therefore, sensu stricto describes phonetically conditioned morphotonemic alternations at the juncture of words or morphemes. Over time, however, tone sandhi has been extended to cover a number of related phenomena, including allotonic variations, intonational effects, and morphologically or syntactically conditioned tone changes. Such a broad and loose usage of the term has one advantage over any strict definition, as Andersen (1986:2) argues in a different context, in that it gives us a convenient label without prejudging the issues involved.

Even though the tonal/registral systems and their historical developments have been investigated extensively in such language groups as Tibeto-Burman, Miao-Yao, Kam-Tai, and Mon-Khmer, tone sandhi has not received as much attention – with the notable exception of the Sinitic group.