Transitive nominals in Old Avestan.

Author:Lowe, John J.

    Old Avestan is widely agreed to attest a variety of adjectives and nouns that optionally or obligatorily display "verbal" government of accusative case "objects"; the same phenomenon (superficially, at least) is found also in other old Indo-Iranian and Indo-European languages such as Sanskrit, Old Persian, Germanic, Latin, and Ancient Greek. Old Avestan grammars contain only lists of forms, or cursory mentions: Reichelt (1909: 230-31) merely provides a list of forms under the heading "Der Akkusativ bei verbalen Nominibus"; Kellens and Pirart (1988-1991: vol. 2: 3-52) have detailed sections on "La syntaxe des cas," but treat all cases except the accusative; West (2011) merely notes the relevant forms, with reference to "verbal rection."

    From the perspective of formal linguistics, however, the existence of such "transitive nominals" is problematic; in fact some influential models of linguistic categorization exclude the possibility. It is therefore necessary to reconsider the evidence of these old Indo-European languages, both for their bearing on formal linguistic models and for a more detailed and coherent analysis of the phenomena in individual languages.

    This paper concentrates on the evidence of Old Avestan exclusively. A full comparative treatment of all the Old Indo-Iranian and Indo-European evidence is, of course, a desideratum; that can only follow, however, synchronic assessments of the distinct linguistic bodies concerned. This paper is a first step towards the latter goal. I have chosen to focus on Old Avestan for two main reasons. Firstly, the corpus is relatively limited, such that it is possible to consider every single alleged instance of a transitive nominal. Secondly, due to the nature of the Old Avestan texts and our still restricted understanding of their language, there is a high degree of disagreement over the precise inventory of transitive nominals in Old Avestan. For this reason, a detailed investigation of the phenomenon as a whole is also of distinct benefit for our understanding of the Old Avestan language and our interpretation of the texts.

    In this paper, I intentionally treat Old Avestan separately from Younger Avestan. It is not only the case that Younger Avestan represents a later stage of a very closely related but nevertheless distinct dialect from the language of the Old Avestan texts, but in addition the language of the Younger Avestan texts was potentially influenced by (a possibly in some respects mistaken understanding of) the Old Avestan language. That a particular morphological category regularly displays transitivity in Younger Avestan cannot be used to support the transitivity of corresponding forms in Old Avestan as long as we cannot be sure that the former is not influenced by the latter. (1)

    The broad inventory of Old Avestan nominals to which transitivity has been assigned may give the impression that almost any noun or adjective with the relevant meaning could have governed an accusative "object." However many of the forms involved are considered transitive by only one or two recent translators. No two translations agree precisely on the inventory of forms to which transitivity must be assigned; a consideration of the phenomenon as a whole, however, enables us to delimit its extent more accurately. It is hoped that, by attempting the latter, some advance may be made in understanding what is and is not a valid way of interpreting particular Gathic passages. For this paper I considered all nominals analyzed as transitive by at least one of the following translators: Bartholomae (1904, =AiWb; 1905, =B05), Humbach (1959, =H59), Insler (1975, =175), Kellens and Pirart (1988-1991, =KP88), Humbach (1991, =H91), Humbach and Faiss (2010, =HF10), West (2010, =W10; 2011, =W11); and for the Yasna Haptarjhaiti also Hintze (2007, =Hi07) and Narten (1986, =N86) (2) In my analyses of the forms I aim to avoid, as far as possible, the appearance of circularity by utilizing the majority verdict of previous scholarship (insofar as such a majority exists), rather than simply following my own preferred interpretation of the texts.

    In [section] 2 I define what I mean by transitive nominals and discuss the problems with formulating such a definition. In [section] 3 I then go through all the Old Avestan evidence for transitive nominals and show that they can be subdivided into several distinct categories. Finally, in [section][section] 4-5 I discuss the consequences of my analysis and draw conclusions.


    My concern in this paper is nominals, i.e., nouns or adjectives, that appear with object-like dependents in the accusative case, that is in a configuration traditionally associated with the verbal system and called "transitive." Cross-linguistically such "transitive nominals" are rare, and their existence is often ignored or explicitly excluded in theoretical linguistic definitions of "objects" and even of "nouns" and "adjectives." (3)

    Non-finite verbal categories like participles and infinitives are not included under the label "transitive nominals" here: although participles, for example, regularly govern objects despite being morphologically adjectives, their ability to do this is entirely dependent on their status as (non-finite) verb forms (cf. Lowe forthcoming on this for Rgvedic Sanskrit). That is, if the verbal stem to which a participle or infinitive is formed is transitive, the participle/infinitive will be also. The transitivity of such verbal forms is cross-linguistically unremarkable since the forms are, in a sense, underlyingly verbs; in contrast it is the existence of transitive lexical nouns and adjectives that is typologically rare and worthy of further investigation. It is, admittedly, not always possible to clearly distinguish non-finite verbal categories from nonverbal derived nominal categories, and in reality non-finite verbal systems tend to consist of a variety of nominal categories that show different degrees of morphological and syntactic integration into the verbal system; in fact it will be seen below that many of the "transitive nominals" of Old Avestan do show a limited degree of morphological association with the verbal system. Nevertheless, it is usually possible to distinguish a set of core non-finite verbal categories, any transitivity of which is uncontroversially dependent on their verbal status; in Old Avestan, this includes, for example, participles in -nt- and infinitives in -diiai. It is not such categories that are of interest here. In this paper I follow the consensus of modern scholarship, wherever possible, in the analysis of a particular form as part of a nonfinite verbal category (participle or infinitive) or as an independent morphological category. (4)

    Cross-linguistically, transitive adjectives are somewhat less rare than transitive nouns, being found for example in a number of Germanic languages (Vincent and Borjars 2010). This is understandable insofar as adjectives are somewhat closer to verbs on the "cline of verbality" between nouns and verbs. Many of the supposed Old Avestan transitive nominals are in fact nouns, which makes them of particular interest from a comparative linguistic perspective. (5)

    Before considering the data, it is worth reviewing and clarifying precisely what we mean when we talk about transitivity and objects in the nominal domain (and more widely). For example Sleeman (1993) and Sleeman and Verheugd (1998) use the term "transitive adjective" for any adjective with a complement, but that is not what we are talking about here: only complements assigned a verbal object case are "objects." There are at least three problems involved in defining the issue at hand: transitivity itself, the multiple uses of different cases, and the different types of verbal government. Limitations of space prevent a full discussion of all these issues, but the major difficulties can be briefly illustrated.

    In Old Avestan, as in many old Indo-European languages, objects are not necessarily restricted to the accusative case. Verbs of ruling, for example, take genitive objects; verbs of fearing may appear with the ablative. However, it is often difficult to determine what kind of dependent we are dealing with, complement or adjunct, especially in the case of oblique case NPs. In Modern English it is at least relatively clear which oblique case (i.e., prepositional construction) dependents are subcategorized for, since the verb in question will be ungrammatical without it. In Avestan, on the other hand, not only do we have only a limited corpus of data and no native speaker judgments on the grammaticality or otherwise of unattested constructions, but we find a considerable freedom for null or ellipsed obligatory complements.

    Perhaps partly for these reasons, the assignment by previous authors of "transitivity" to nominals in Old Avestan is almost entirely restricted to those that appear to govern the accusative case. Nominals can, of course, appear with dependent nouns in other cases; most importantly, semantic "objects" usually appear in the "objective genitive," but any other case relation is possible. The problem is how to distinguish non-accusative complements from non-accusative adjuncts. For example, it would be theoretically possible to interpret the genitive often found with the noun xsa[??]ra- 'rule, power' as an "object" in the corresponding case as the object of the related finite verb. (6)

    (1) yueam zauuistiianho / iso xsa[??]ramca sauuarjham (Y 28.9c)

    H91: You (are) the swiftest invigorations and the power over benefits.

    But there is no reason not to take this merely as an objective genitive adjunct. The same is true of other oblique cases. So, W11: 14 takes the noun zrazdaiti- 'trust' at 43.11 (2) as "verbally" governing the locative, but it might just as, if not more, easily be treated as a simple locative adjunct.

    (2) sadra moi sas / masiiaesu + zrazdaitis (Y 43.11de)


To continue reading