Passive or Intransitive? The case of hark- 'perish' and voice / transitivity in Hittite
The paper discusses the use of intransitive verbs as passives in Hittite. In particular, it offers a new assessment of a number of uses of the verb harkw hich is traditionally analyzed as intransitive 'perish'. The careful study of the contexts as well as cross-linguistic parallels suggest that in a number of cases the verb functions as passive to the two verbs that are derived from the same root - harnink- and harganu-, both meaning 'destroy'. More generally, the paper argues that any intransitive verb in Hittite can potentially be used as passive if there is a transitive verb from the same stem, root or even just a semantically compatible verb. This happens irrespective of whether the intransitive verb is active or middle and holds true for the New Hittite period. In this way the paper presents a single account of three classes of verbs: suppletive passives (akk- 'be killed' alongside kuen- 'kill'), intransitive verbs alongside a derived causative (hark- 'be destroyed' alongside harganu- / harnink- 'destroy') and middles alongside active verbs (šarra- 'be divided' alongside šarra- 'divide') in Hittite.
In the broader cross-linguistic perspective the Hittite data attest the standard alternation anticausative — causative from the same stem with anticausative attesting a supplementary passive function for the causative verb, also common in the languages of the world. The curious fact that one of the passives for the causative verbs harnink- and harganu- derived from the anticausative hark- is supplied by the anticausative hark- itself is important evidence in favor of the indistinguishability of anticausatives and passives and an argument to divorce passive from the formally marked middle voice or from analytical forms.
Hittite is a language with morphological marking of voice as middle or analytical passives. However, voice in Hittite is closely tied in with anticausatives. It was shown by [Luraghi 2010; 2012; Inglese 2018] that passive voice is diachronically a reanalysis of anticausative and synchronically in one of the periods of the Hittite language evolution it cooccurs with anticausatives. I extend the proposal to the broader domain and show that passive in Hittite is not associated with any formal voice marking, but rather with anticausatives generally. Anticausatives marked both by active verbal morphology and by middle verbal morphology (both synthetically and analytically) can occasionally function as passives to the alternating causative verb.