Theiling Online    Sitemap    Conlang Mailing List HQ   

Re: Markedness of passives and antipassives

From:Alex Fink <a4pq1injbok_0@...>
Date:Thursday, April 5, 2007, 1:28
On Fri, 23 Mar 2007 15:26:34 -0500, Eric Christopherson <rakko@...>

>Somewhere, I seem to have picked up the idea that passive voice verbs >are marked and therefore do not occur as frequently or as "naturally" >in a given language as active verbs. I would like the community's >thoughts on how true this is. > >One thing that makes me question this received wisdom is that I >remember reading somewhere about a language where passive >constructions actually do, contrary to expectation, occur in a >proportion similar to that of actives. Does anyone know what language >that might be (or of other such languages)?
Ma:ori is apparently one such language, where passives may even occur more than actives. The everything_2 article ( says | Maori loves passive verbs and uses them more often than active. Especially in | circumstances of obligation or necessity, the passive is most likely to be used. As Roger says, "trigger" languages are other examples, if object focus is really well analysed as a passive.
>Another reason is that there are some languages where passive >constructions, at least in certain contexts, evolve diachronically >into active ones; for instance, the ergative past tense constructions >in a lot of modern Indic languages. If passives are marked and >perceived as unnatural, how did they take over for the ancestral, >active, past tense paradigms? Were there sound changes or analogy or >something that destroyed the original past constructions? > >Finally, are the same issues (markedness and naturalness) present >with antipassive constructions in ergative languages? I'm guessing >they are. The main reason I ask is that I'm working from an ergative >conlang and I would like to generalize antipassive forms, with the >result being accusative constructions.
I think, although sound change etc. could be precipitating factors, they need not be so in any given case, and there are generally deeper reasons for this sort of change in structure of grammatical relations. I suppose this is a matter of naturalness; I'm not sure how it relates to markedness. There's a semantic connection between the past and perfect aspects and the passive, which goes roughly like this. Sequences of events which are reported in the past and/or perfective are commonly used to discuss a common participant to which they all happened: it's this participant whose present, perceptible state is the most affected by these events, as opposed to (say) a common instigator or volitional participant. This most affected participant is the patient of transitive verbs, and of course the sole argument of intransitives. So, in such a sequence of clauses, a language is best served by an S/O pivot treating intransitive subject and transitive patient the same, or at least a way to manoeuvre S and O into the same syntactic position. For an accusative language, this means the passive. So this motivates the appearance of split ergativity in a nominative lang, with the ergative patterns appearing (perhaps from old passives) in past or perfective clauses. There's a corresponding connection motivating antipassives; when these are favoredly selected by some TAM, it'll be a future or a purposive construction. Here the common thread is that the participant most likely to control a future or purposive event is the one with the most volition, i.e. the intransitive subject or transitive agent, so that an S/A pivot or systematic use of antipassives is useful. So it's certainly possible for antipassives to generalize, first to such clauses and thence to others. Dixon cites Mayan, Tibeto-Burman, the Australian langs, and South Caucasian and Iranian as families with ancestors displaying ergative features in which this sort of thing has happened to varying extents. There are of course other factors which correlate with accusative vs. ergative marking. One of these is position on the nominal hierarchy, ranging from nouns most likely to be A (speech act participants) to those most likely to be O (inanimates); here the impulse behind a split would be that it's only the less likely syntactic-semantic relation which needs to be marked for a given noun. (Most of what I've said here is from Dixon's _Ergativity_, in the Cambridge Studies in Linguistics series.) Alex