The Indo-European language family contains many 'small words' with various adverbial meanings and functions, including preverbs. The term 'preverb' is used to label any of a variety of modifying morphemes that form a close semantic unit with a verb, including both words and prefixes (Booij and Kemenade 2003). Some Indo-European languages not only have preverbs, but also allow more than one preverb to modify the same verb root. Sanskrit, the focus of the present study, is one such language. The preverbs in Sanskrit 'stack' onto the verb root, such that all of the preverbs alter the meaning of the verb. In Vedic Sanskrit, only two preverbs are ever combined with a single verb (Macdonell 1975), but in Classical Sanskrit, there are also some, admittedly relatively unusual, cases involving three preverbs.
I show that Sanskrit preverbs exhibit consistent tendencies in their relative positioning. For example, the preverb abhi is most usually found in the position farther from the verb, where the preverb parā is categorically found next to the verb. It has long been believed that where Sanskrit showed multiple preverbs, the ordering of those elements was determined primarily by “the requirements of the meaning” (Whitney 1925: §1080), but I provide numerous counterexamples to this claim. There are many cases where the order does not influence the meaning.
There is considerable evidence showing that, rather than being explainable in purely synchronic terms, e.g. based on phonological or semantic properties, preverb ordering instead reflects properties inherited from Proto-Indo-European. I base this claim on two findings. First, my data show strong correlation between ordering properties of preverbs in Classical Sanskrit and those found in the older Vedic language, such that the classical situation is clearly a ‘crystallization’ of the older Vedic patterns. Secondly, I have determined that there is a relationship between ordering properties of Sanskrit preverbs and the ordering properties of cognate preverbs in other Indo-European languages that allow preverb stacking.
While a large number of IE languages have preverbs, a much smaller number also allow stacking. Greek, Irish, and various Slavic languages are well known for this; Irish is particularly notorious for preverb stacking. According to Thurneysen, as many as five preverbs may 'occasionally be found'. This is many more than are ever combined in Sanskrit, even in the Classical language. However, Kim McCone (1987) has presented a relative ordering hierarchy for Old Irish. By comparing my data with information presented in McCone, I show a strong correlation between the ordering trends of Sanskrit preverbs and those of the Irish verbal prefixes.
Additionally, there is a correlation in ordering between certain Greek prepositions and those of both Irish and Sanskrit. This is further evidence that Classical Sanskrit preverb ordering trends are indeed inherited rather than synchronically generated. Particularly, this similarity between the three languages shows that rather than simply being inherited from earlier Indic, as the close similarity between the Vedic and Classical data suggests, these trends are possibly inherited directly from Proto-Indo-European.