Vedic possesses two different forms in the nominative plural of a-stems (e.g. áśva- ‘horse’): one in -ās (áśvās ‘horses’), and one in -āsas (áśvāsas ‘horses’). Cognates of these two variants in the nominative plural are also found in Avestan and Old Persian (Av. -ā̊ : OP -ā : Ved. -ās vs. Av. -ā̊ŋhō : OP -āha : Ved. -āsas), indicating that this variation was already present in Proto-Indo-Aryan.
The goal of this project is to study the original distribution of these forms. To this end, primarily Old Indic data is consulted, as the Old Iranian text corpus is not sizeable enough to supply sufficient data relevant to the study of this variation. Based on a preliminary survey of the data, the following working hypotheses are posited: (i) The criterion expressed by the opposition of -ās vs. -āsas applies to the entire nominal phrase, as only one element of a given nominal phrase may show this long ending -āsas. (ii) The function of the two forms is to designate the position of nouns on a scale of agentivity: -āsas is used to indicate a high degree of agentivity, whereas -ās indicates a lesser degree. (iii) The long form -āsas can be traced back to a univerbation of the original short form (Ved. -ās < PIE *-ōs) with the nominative plural of the Indo-European reflexive pronoun *=s-es.
Should the hypotheses (i) – (iii) prove correct, we could assume an opposition such as *sutó-es ʻjuicesʼ (>> Ved. sutā́s ʻjuicesʼ) vs. *sutó-es=ses ʻjuices themselvesʼ (>> Ved. sutā́sas ʻjuicesʼ) for early Proto-Indo-European. The element *=s-es may have had intensifying function, as in English “John himself opened the meeting.” Under this assumption, it would not be at all unlikely, from a typological perspective, that such an element would develop into a marker of high agentivity.