Acoustic correlates of the voicing contrast in Lebanese Arabic singleton and geminate stops

This study explores which acoustic correlates best distinguish the voicing contrast in Lebanese Arabic, a language with a two-way voicing contrast that occurs with both singleton and geminate stops. The required timing, phonation and articulatory strength settings for each contrast act synergistically in the voiceless set, but it is unclear how the contrasting requirements for voiced geminates are implemented. Twenty adult speakers were recorded producing target words with medial singleton and geminate stops preceded by long and short vowels. Several temporal and non-temporal measures (duration, VOT, percent voicing, f0, F1, intensity, H1*-H2*) were taken in the surrounding vowels and in the closure and release phases. Results show that closure duration is the most important cue for distinguishing both voicing and gemination. Active and passive voicing patterns in the closure of voiceless and voiced stops point to [voice] as the main distinctive feature, with [tense] as a secondary feature for voiceless and for geminate stops, with a graded effect. Non-temporal correlates show geminates to have increased tension and creak. Crucially though, voicing is still active in voiced geminates, and release properties have more in common with lenis than fortis languages, leading to a complex profile for this marked category of sounds. (C) 2018 The Authors. Published by Elsevier Ltd.