Turns out final devoicing isn’t actually final devoicing

Dutch people pronounce “dog” as “dock” and “bed” as “bet” and I had always learned that this was called “final devoicing”. But when I started researching this phenomenon properly for my master’s thesis, boy did I fall into a rabbit hole.

(It’s not only Dutch people that do this, by the way, but also Germans, Poles, Russians, Catalonians, Lithuanians and more.)

First lets do a quick review of what voicing is: a voiced phoneme is when the vocal cords are vibrating while you are making the sound. Touch your throat while you say “sssssss” and then “zzzzzz”. You will feel a vibration for “zzzzz” but not for “ssssss” – that’s the vocal cords vibrating. When they vibrate, we call the sound voiced, when they don’t, it’s voiceless. Final devoicing, then, is when a sound at the end of a word that should be voiced, like “eyes” (which is pronounced as eyezzzzz) comes out voiceless, like “ice”.

So far so good. However (this is the entrance to the rabbit hole) it turns out that, like so many things in life, context is everything. Take these three sentences:

“I opened my eyes.”

“I kept my eyes closed”

“I kept my eyes open.”

In the first sentence, the /z/ sound is followed by a pause, in the second, the /z/ is followed by a voiceless sound /k/ and in the third the /z/ is followed by a voiced sound /əʊ/ (vowels are always voiced).

It turns out that native speakers usually do NOT voice the last sound in “eyes”. They only make a voiced /z/ sound at the end of “eyes” when it is followed by a voiced sound, and even then, only a third of them do! The other third does something called “partial devoicing” which in this case means the very beginning of the /z/ sound is voiced but the rest of it isn’t, and the other third doesn’t have any voicing at all.

It happens for plosives (/d/, /b/ and /g/), too, though the effect is less striking.

So if final devoicing is not final devoicing, what the heck is it?

When a native speaker says “I opened my eyes”, it is clear that the last word is “eyes” and not “ice”, even though the last sound is actually a voiceless /s/. So what’s going on?

First thing to know is that apart from voiced and voiceless, you can also divide these kinds of obstruents* into fortis and lenis. Fortis obstruents are the voiceless ones, such as the /s/ in “ice”, which are called fortis because they are made with a lot of force, i.e. a strong burst of air. Lenis sounds are ones like the /z/ in eyes, but in this case we’re not talking about the vocal cords vibrating, but we’re only talking about the fact that the airflow isn’t as forceful as it is with lenis consonants. So that explains the graphs above where the /z/ turned out to be pronounced in a voiceless way so often: the vocal cords aren’t vibrating, so it’s not voiced – but it is lenis. This means that even though the /z/ in “eyes” isn’t voiced, it still sounds different to the /s/ in “ice”. It’s pronounced more softly, with a milder burst of air.

*obstruents are plosives such as [p, t, k, b, d, ɡ], fricatives such as [f, s, v, z] (and affricates, but let’s leave those out for now). There’s also another type of consonant, namely the sonorants, like [m, n, l, r]. These are always voiced, and fortis or lenis doesn’t apply to them – you can’t make the [m] sound different by putting more force into your puff of air (if you are like me, you will now try this out, and sound very funny in the process 😉 )

The fortis/lenis thing means that we shouldn’t actually talk about final devoicing when we talk about Dutch people pronouncing “dog” as “dock” and “bed” as “bet”, because the voicing doesn’t really matter. The phenomenon is more properly described as “fortis/lenis neutralisation“, and during my literature research I saw that newer studies indeed refer to it this way.

Interestingly, the Dutch word for final devoicing is “eindklankverscherping” and the German word, which I also sometimes see used as an international word, is “Auslautverhärtung“. These two terms are correct representations of what is actually happening. I’m so used to using English terminology for everything that it hadn’t even clicked with me that these words were saying something different to the English term, and I spent all this time getting confused about something that I could have just understood in one go if I had been doing my research in Dutch. Serves me right for not putting more trust in good old Dutch!

But wait, there’s also vowel length

So is that it? Should we stop teaching Dutch people about voicing and start teaching them about forceful air flow? Well, yes, but that’s not the whole story. There’s another reason why “dog” and “dock” sound different to native speakers, which according to research is actually more important: the vowel length. The /o/ in “dock” is much shorter than the /o/ in dog for native speakers, but not for Dutch people. Dutch people make these vowels the same length, and according to the books and articles that I’ve read, that’s actually what is causing the confusion.

So what should we be telling Dutch people to get them to stop saying “dock” and start saying “dog”

It would seem then that the best advice is to make the vowel a bit longer and the last sound a bit softer. “Doooock”. Something like that. I have tried this out on exactly one test subject, my boyfriend, who was the main reason I started researching this thing in the first place, and I have to say it has really made a difference. But he’s just one person, so that’s just anecdotal evidence. For my master’s thesis, I’ll be asking more Dutch people if this is the advice that makes a difference, and I’ll be giving it a slight twist. I’ll post updates on this website, so watch this space!

I have already asked 137 English teachers what they do to help their students. It turns out that most just make sure their own pronunciation is correct and that they address it when they hear their students make this kind of pronunciation error, but they don’t do much else. Of the 28% of teachers that tell their students explicitly about final devoicing (which we now know is not in fact final devoicing), 55% talk about vibrating vocal cords – which, as we also now know, is not actually very good advice. 50% give the advice to make the vowel longer, and I don’t know how many teachers advise to make the last consonant softer because unfortunately when I made the questionnaire I hadn’t figured that bit out yet so I didn’t ask. I did ask some Dutch people about tips they had had in their life to combat this pronunciation issue and a fair few of them did answer “make the last consonant softer” so it does seem to be advice that exists in the world.

I’d love your input!

If you are reading this and thinking “what is this woman talking about, the best piece of advice to give someone is clearly …” then by all means, let me know in the comments!


The graphs are from Gonet (2012), and a great explanation of the fortis/ lenis and vowel thing can be found in the excellent master’s thesis by Anne van Leeuwen (2011).

Gonet, W. (2012). Voicing of word-final obstruent in English connected speech. Speech and Language Technology, 14/15, 181–190. Retrieved from https://www.academia.edu/download/47195689/Gonet_-_From_Jassem_Volume_-_Voicing_of_WF_Obstruent.pdf

van Leeuwen, A. (2011). Mastering the temporal pattern of English. Retrieved from https://dspace.library.uu.nl/bitstream/handle/1874/210344/thesis_annevanleeuwen.pdf

Heddwen Newton is a teacher and translator. Her website EnglishforDutchpeople.nl is about efficient and unconventional ways for Dutch people to improve their already good English, and other nerdy stuff to do with English and Dutch. She also owns the Dutch website HoezegjeinhetEngels.nl where she discusses difficult-to-translate Dutch words and their least-bad English translations.

Leave a Reply

Your email address will not be published.