I'm native speaker of Georgian, a language which has phonemic distinction between voiceless, 'voiced', and ejective consonants (according to study and Wikipedia)

though when repeating Georgian voiced stops to a speaker of a language with voiceless and voiced unaspirated stops They mistake my voiced stops for their voiceless unaspirated plosives for some reason even to my ears they sound identical to IPA [b d ɡ].

2 Answers 2


The first thing you should do is check the assumption that either language has voiceless and voiced unaspirated stops. French is an example of such a language, so if the person you're working with is a native speaker of French, that is different from a native speaker of English. There isn't a great disparity between phonological and phonemic analysis in French, as there is in English. The expectation that /t/ sounds like [t] and /d/ sounds like [d] is only valid if the stimulus sounds and listener native language sounds are in fact identical. Otherwise, you have to engage in a comparison of acoustic properties of the various sounds, to see which Georgian sounds are closest to a particular other-language sound. Duration, release bursts, spectral tilt and F0 effects on the following vowel as well as the timing-pattern of vocal fold vibration (such as VOT) are relevant acoustic cues. Previous literature on Georgian phonetics suggests that supposedly voiced stops are more like those of English and less like those of French, that is, they are not fully voiced. Butskhrikidze concludes, correctly IMO, that the matter of how to treat the two non-ejective consonant series is unresolved, and needs experimental investigation in order to justify the claim that there is a voicing contrast versus an aspiration contrast.


The acoustic difference between voiced, tenuis (voiceless unaspirated), and aspirated stops mostly comes down to voice onset time: the difference in time between the release of the stop and the start of voicing. If voicing starts before the stop is released, it's more "voiced". If it starts significantly later, it's more "aspirated".

Note though that this is a continuous value rather than a discrete one. In other words, you can have an extremely voiced stop, or a slightly voiced stop, or anywhere in between. Different languages draw boundaries in different places on this continuum to break it up into discrete categories. For example, English and Spanish both have a two-way distinction: stops that fall on one side of the dividing line are "voiced", and stops on the other side are "voiceless". But the boundary line for English is more on the "aspirated" side, and the boundary line for Spanish more on the "voiced" side, meaning that quite a lot of English "voiced" stops can sound "voiceless" to Spanish-speakers and vice versa. (For this reason, some phonologists prefer to use the terms "lenis" and "fortis", to make it clear that they're talking about a language-specific phonological boundary rather than something universal.)

Presumably the same thing is happening here. Stops that fall into the category of "voiced" for you, fall into the category of "tenuis" for your listener, since the boundaries are drawn in different places in their native language. Their "voiced" stops are probably even more voiced, with even more voicing happening before the release of the stop.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.