Many conflicting ideas have been put forward concerning the syllabification of Vedic Sanskrit. Varma (1929:61-86) presents the statements of the Prâtis"âkhyas, ancient Indian treatises devoted to the correct pronunciation of the various Vedas. All agree that consonants at the edge of words go with their adjacent vowels, and that a single consonant between vowels goes with the following syllable. But they disagree on how to treat clusters of consonants. Most of the Prâtis"âkhyas state that the consonants split up (Varma 1929:62). Finally, the Taittirîyaprâtis"âkhya draws a distinction: most consonants are split into two syllables, but they are kept together if the second is an approximant (21.7, apud Varma, p. 75), or if the sequence is a stop followed by a fricative (21.0, apud Varma, p. 71).(2) This table gives some examples of how the various schools would syllabify some of the words found in the first hymn of the R.gveda. It should be kept in mind that the discussions in those treatises are very cursory, and that the authors may not have intended these precise outcomes. The statement of the R.kprâtis"âkhya that either syllabification is permitted may originally have meant that the alternation is determined in part by some unspecified property of the word, not necessarily entirely by the whim of the speaker. More importantly, the authors of the Prâtis"âkhyas are only concerned with fixing the pronunciation of a liturgical text composed centuries before their time. They did not put their definition of the syllable to any phonological or grammatical use. If one is inclined to believe that syllables are ultimately phonological or that there can be differences between phonetic and phonemic syllables, then the statements of these phonetic treatises should be approached with caution.
Thus already in ancient India there were two fundamentally different approaches to syllabification of intervocalic clusters. The first I shall refer to as the cluster-splitting approach (CS)onset-maximization (OM), which decrees that as many consonants syllabify with the second vowel as is compatible with general constraints on the structure of the onset. Different linguists have formulated different onset structure constraints. According to the theory of the Taittirîyaprâtis"âkhya, those constraints are that a consonant may be followed by an approximant, or that a stop may be followed by a fricative.
The same two classes of approaches have continued into more recent times. Most Indo-Europeanists subscribe to the CS theory, that all clusters split in Vedic Sanskrit (Hermann 1923:261, Vaux 1992). In recent years, theoretical linguists have published several proposals taking an OM approach, that clusters syllabify together in an onset if they can (Rice 1990:309-311, Cho 1990:204; Kessler 1993 argued this for Classical Sanskrit, a later variety of the language). They differ among themselves only in certain small details concerning what the possible onset clusters are. Both sides of the debate have strong evidence on their side. In the following sections I shall briefly present the evidence that I am aware of for each analysis, and then show that ambisyllabicity would account for both sets of data.
[Forward to Data from the Rigveda: Cluster splitting: Metre] [Up to Background] [Up to table of contents]