Voice 257
(23) a. Ale' e-kekke' bari', bi' embi' rowa.
yngr.sibling OV-bite yesterday by goat that
‘That goat bit Little Brother yesterday.’
b. *Ale' ekekke' bari' embi' rowa.
(24) a. So polisi maleng rowa e-tangkep.
by police thief that OV-catch
‘The police caught the thief.’
b. *Polisi maleng rowa etangkep.
As stated above in section 1, despite the fact that word order and gram-
matical relations seem to align this structure with passives in other languages,
the functional load of object voice in Madurese (and other related languages)
differs from passives in other languages. Similar to passive, object voice is often
used when the agent is unspecified, as in (25).
(25) Ale' e-kekke'.
yngr.sibling OV-bite
‘Little Brother was bitten.’
(26) Maleng rowa e-tangkep.
thief that OV-catch
‘That thief was caught.’
But distributionally, the ratio of object voice structures in a discourse is far
higher than the ratio of passives in the relevant languages. In Madurese narra-
tives, approximately 50-60% of all voice marking on transitive predicates is
object voice, as reported in Davies 2005 and discussed in more detail in Chapter
10 section 4. In English, the percentage is considerably smaller. For example,
Blankenship (1962) reports that passive clauses made up only 7.8% of clauses
in oral and 17.9% in written discourse; other reports based on experiment
(Poole and Field 1976, Biber 1988) and casual observation (e.g. Quirk et al.
1985) also assert passive has a low frequency in English. And similar reports
are available for other languages, e.g. Chinese (Xiao, McEnery and Qiang
2006). Additionally, in languages such as English, the percentage of passives
clauses with overt agents is quite low: 13-20% of passives have an agent phrase
(Svartik 1966, Duškova 1971, Givon 1979) and for German passives 17.8% has
been reported (Stein 1979). For Madurese, 60-70% or more object voice clauses