The timing of geminate consonants

In current phonological theory, geminate consonants are distinguished from non-geminates by a difference in units on the timing tier in an autosegmental representation, other features being the same; and while there are geminates of three kinds according to how they arise (tautomorphemic, concatenated, and derived by total assimilation), the representation of all three on the timing tier is the same. The present paper investigates two questions: what exactly is the acoustic difference between geminate and non-geminate consonants (focusing on voiceless stops); and are there acoustic differences between geminates derived from different sources phonologically. We examined two unrelated languages, Bengali and Turkish, and found that the single overriding cue distinguishing geminate from non-geminate stops is the duration of the closure. In Bengali, we examined geminate stops derived from the three different sources and found no significant difference in closure duration. Thus the autosegmental representation of the difference between geminate and non-geminate consonants in terms of a timing difference is vindicated, and we have an acoustic correlate for the timing measure. As a secondary result, we have shown that total assimilation leading to the formation of a geminate constitutes an instance of neutralization that is phonetically complete.