The Sanskrit data constitute a strong case for the existence of ambisyllabicity. They cannot be explained away by some of the alternatives presented to Kahn's data for English. Kiparsky's invocation of stress feet (1982) cannot explain situations such as the formation of perfects, which is sensitive to both syllabifications simultaneously. The fact that this ambisyllabicity cannot be connected with the Germanic data in terms either of areal contact or of historically antecedent structures (the breakdown of contrastive gemination in the Germanic cases) bolsters the case for ambisyllabicity's being a general linguistic phenomenon. It is also encouraging that this ambisyllabicity eventually developed a phonetic correlate (length), which proves that the bisyllabic structures were a living part of the language, as opposed to the possibility that the maximal onset patterns found in the morphology were frozen remnants of an earlier phonology.
In addition, the fact that ambisyllabicity is associated with gemination both in Germanic (gemination develops in ambisyllabicity) and in Sanskrit (ambisyllabicity develops into gemination) bolsters the case for their having very similar bisyllabic representations. Indeed, the fact that ambisyllabicity and gemination never contrast in Sanskrit even though both are present in the language strongly supports the conjecture of Borowsky et al. (1984) that the two phenomena have identical phonological representations, distinguished only at a phonetic level. Although Suzuki (1985) and Borowsky et al.