Corpus-based approach meets LFG: The puzzling case of voice alternations of kena-verbs in Indonesian

The published, revised, and peer-reviewed version of this paper goes with the following title:

Data, codes, and figures for the quantitative analyses can be accessed from here or here.

In this talk, we show that voice alternation is not only about meaning-preserving phenomenon and re-alignment of grammatical relations and semantic roles. We argue, instead, that such phenomenon also involves lexical-meaning construction with holistic morpho-semantic and usage properties that may differ between voice types for a given verbal root.

We contrast the distribution of different senses for verbs based on the Indonesian verbal root kena 'get hit' in different voice constructions (i.e. dikenai, dikenakan, mengenai, and mengenakan).

Based on quantitative, corpus-based evidence, we show that different meanings conveyed by a given verb (e.g. kenai) find different expression in different voice types (e.g. the different predominant sense for the A(ctive) V(oice) mengenai vs. PASS(ive) voice dikenai). Another important finding is that certain sense of a given verb can be directly constructed in one voice type (i.e. PASS) without any usage evidence of that sense in AV (hence, no voice alternation for that sense, let alone its meaning-preserving status).

Our findings challenge the status-quo in linguistic theorising that voice alternation is a matter of re-alignment of syntactic function and semantic-role as well as meaning-preserving phenomenon (cf. Kroeger 2005: 271); however, from the usage-based perspective, voice in Indonesian can be seen as morphological constructions as they may involve lexical-meaning construction for each voice type for a given root, bringing about distinct usage/semantic preferences between alternating voice (e.g. between AV and PASS form). Moreover, we argue that voice alternation needs to be relativised to (i) a certain (rather than all potential) sense(s) of a verb, and (ii) (statistical) usage constraints of the verb's semantics in certain voice.

The precursor of this study can be seen in G.P.W. Rajeg, I M. Rajeg, I W. Arka (2020).

We are also working on language-production experimental data to further test our corpus-based findings regarding the predominant sense for a given verb in a given voice.

Rajeg, Gede Primahadi Wijaya, I Made Rajeg & I Wayan Arka. 2020. Corpus-based approach meets LFG: Puzzling voice alternation in Indonesian. OSF. doi: 10.17605/OSF.IO/YMD2V. osf.io/ymd2v.