Devoicing

Hi everybody!

I'm a bit confused: do we have "final consonants devoicing" in English??
I thought the rule goes like this: devoicing - after voiceless consonants.
no devoicing - after voiced consonants and vowels.

Right?
What about lease [li:s] - leased [li:st] or [li:zd] ???

I think I'm going crazy)))

Thanks!

For regular verb endings for the past simple tense and the past participle (-ed endings):

1. If the last _sound_ in the base form is /t/ or /d/, the -ed ending is pronounced as /ɪd/.

2. If the last sound in the base form is a vowel or voiced consonant and not /t/ or /d/, the ending is pronounced as /d/.

3. If the last sound in the base form is an unvoiced consonant and not /t/ or /d/, the ending is pronounced /t/.

These are the rules for standard pronunciations (GAE and RP, mainly). Some other pronunciations may differ.

So, then it is [li:st], isn't it?

Thanks a lot!

I thought I'm crazy. But I'm not! ))

>>Hi everybody!

I'm a bit confused: do we have "final consonants devoicing" in English??
I thought the rule goes like this: devoicing - after voiceless consonants.
no devoicing - after voiced consonants and vowels.<<

The direction of assimilation when it comes to consonant clusters within English dialects varies from dialect to dialect and case by case. For example, with words like "leased" in my dialect, the assimilation is progressive (that is, affects following phones), whereas in words like "have to" and "need to" the same assimilation is retrograde (that is, affects preceding phones); in these cases, the assimilation in question takes the form of devoicing.

As for *word-final* devoicing, such is just a weird feature of my own dialect which is reflected in my transcriptions here, and which is in no means at all usual for an English dialect. Not quite the same as such in German, Dutch, Polish, or Russian, due to its preserving a fortis-lenis distinction and its marking fortis-lenis values in the length of the preceding vowel, but still quite weird for an English dialect.

>>Right?
What about lease [li:s] - leased [li:st] or [li:zd] ???

I think I'm going crazy)))

Thanks!<<

Such would be [list], as the <-ed> ending is always realized as [t] after a word root which ends in an lenis obstruent other than /t/.

A very nice explanation, Travis!
Thank you!

I am a little bit confused as regards what is it that Nadin calls "devoicing of final consonants". I believe that this means that voiced consonants are somewhat devoiced before a pause or some other voiceless segment. That is, the words "lice" and "lies" are phonemically distinct /laIs - laIz/. It would seem that the only difference between them is just the voiced final consonant, but this is not really so: vowel length is of a greater importance. In final position in a tone unit, both /s/ and /z/ might sound identical, yet the vowel will be clipped when you originally had a /s/ but never before /z/.

Yes, as Francisco points out, the vowel length is of great importance in such positions. Many/most English speakers do at least partially devoice consonants at the end of words but the vowels are longer before phonemically voiced consonants. So, even if a /d/ is devoiced to [d_0] word finally, it'll still be different than the word ending in /t/ --> [t].


<bed> /bEd/ --> [bE:d_0]
<bet> /bEt/ --> [bEt]

<bag> /b{g/ --> [b{:g_0]
<back> /ba{k/ --> [b{k]

I should clarify what I meant by word-final devoicing in my dialect (that is, that of the Milwaukee, WI area). Yes, English across the board does have the *half-voicing* of word-final consonants, but it does not have *full* word-final devoicing, where there is no realized constrast in voicing at all between lenis and fortis obstruents in word-final positions. The only exceptions to this is in more "careful" speech where a word is being followed by another word starting with a vowel or a lenis obstruent, or in very emphatic speech. Even though much of North American English at least does appear to have full devoicing of lenis fricatives in certain word-final positions where they are preceded by a sonorant consonant, and where they are also not followed by a vowel or a lenis obstruent, such does not amount to true full word-final devoicing.

>>I am a little bit confused as regards what is it that Nadin calls "devoicing of final consonants". I believe that this means that voiced consonants are somewhat devoiced before a pause or some other voiceless segment. That is, the words "lice" and "lies" are phonemically distinct /laIs - laIz/. It would seem that the only difference between them is just the voiced final consonant, but this is not really so: vowel length is of a greater importance. In final position in a tone unit, both /s/ and /z/ might sound identical, yet the vowel will be clipped when you originally had a /s/ but never before /z/.<<

The situation in my dialect is similar to what you describe, but not quite as simple. For starters, there is not a clear merger of /s/ and /z/, or any other pair of fortis and lenis obstruents, in word-final positions in my dialect despite them not being distinguished voicing-wise, as they still have a clear fortis-lenis distinction, which could be interpreted in terms of a length distinction of sorts. Note though that this fortis-lenis distinction in final positions is far more marked in "careful" speech than it is in most everyday speech and especially in very informal speech, where the actual distinction can be nearly absent.

As for the length allophony, such is present in (most likely) all North American English dialects, but due to the weakening of distinctions between word-final consonants due to full word-final devoicing in my dialect, my dialect has promoted such to a significance greater than that such in most NAE dialects, such that speakers of other dialects, such as even people from areas of Illinois and Indiana, often comment that vowels are often very long, in their perception, in the dialect here.

>>Yes, as Francisco points out, the vowel length is of great importance in such positions. Many/most English speakers do at least partially devoice consonants at the end of words but the vowels are longer before phonemically voiced consonants. So, even if a /d/ is devoiced to [d_0] word finally, it'll still be different than the word ending in /t/ --> [t].


<bed> /bEd/ --> [bE:d_0]
<bet> /bEt/ --> [bEt]

<bag> /b{g/ --> [b{:g_0]
<back> /ba{k/ --> [b{k]<<

The main thing is that what you are describing here is half-voicing, which is common in English dialects, but I myself use the same transcriptions to mark unvoiced lenis obstruents, which are not simply half-voiced, yet which may still contrast with unvoiced fortis obstruents. Unfortunately, neither IPA nor X-SAMPA provide any useful diacritics for marking half-voicing as opposed to true full devoicing proper which still preserves the lenis/fortis status of the phones in question. Normally, though, X_SAMPA _0 and its IPA equivalent seem to mark pure devoicing, and not half-voicing, and hence I still use it here, even though if it did mark half-voicing, then I would likely have to write pairs like those above as:

<bed> /bEd/ -> [bE:t]
<bet> /bEt/ -> [bEt:] (actually normally [bE?] for me)

<bag> /b{g/ -> [b{:k]
<back> /b{k/ -> [b{k:]

Typo of mine above: <back> should be /b{k/ and not /ba{k/.

Travis, another interesting thing I just realized about your consonants is that /t/ and /d/ would still usually be distinguished (besides the lenis-fortis thing) because in that position /t/ --> [?] while /d/ --> [t].

Well, I'm aware of devoicing in everyday speech, or in inaccurate pronunciation. Besides, it is said that American speakers tend to devoicing of most voiced consonants in a final position,like in 'prove it' - [pru:fit], while an Englishman would say [pru:vit].
What I was interested in, was a simple general rule for devocing of consonants word-finally. Like, voicing - after vowels and voiced consonants, devoicing - after all unvoiced, except for /t/.
Thanks to everybody who replied!

<<Besides, it is said that American speakers tend to devoicing of most voiced consonants in a final position,like in 'prove it' - [pru:fit], while an Englishman would say [pru:vit].>>

That definitely doesn't apply to my speech. For me, "prove it" would use [v] and would be clearly different from "proof it".

I should add, just to be clear, that I am from the United States. ;-)

<<Well, I'm aware of devoicing in everyday speech, or in inaccurate pronunciation. Besides, it is said that American speakers tend to devoicing of most voiced consonants in a final position,like in 'prove it' - [pru:fit], while an Englishman would say [pru:vit].>>

I've never heard of that. In fact, "prove it" with [f] doesn't sound right to my ears. It sounds like "proof it" (as in the verb to proof a paper). Partial devoicing of final consonants is common to all/most varieties of English but I would think very few would do so when followed by a vowel such as in "prove it." I definitely have fully voiced [v] there.

<<What I was interested in, was a simple general rule for devocing of consonants word-finally. Like, voicing - after vowels and voiced consonants, devoicing - after all unvoiced, except for /t/. >>

I think Mxsmanic's three-part rule should help you there.

>>Travis, another interesting thing I just realized about your consonants is that /t/ and /d/ would still usually be distinguished (besides the lenis-fortis thing) because in that position /t/ --> [?] while /d/ --> [t].<<

This is except that while my "normal" realization of word-final /t/ after a sonorant is [?], if it is a cluster with another obstruent it is still [t], and it is still in free variation with [t] proper in all places where it may be [?].

Travis, you talk about two degrees of devoicing (or at least, two distinguishable kinds): full and partial devoicing. I find that, as you said, IPA has no special diacritic for this feature and phoneticians are inconsistent with the use of the term and diacritic for devoicing. I am doing some research, but it has been rather unfruitful up to now. Do you know of anybody who tackles this issue thoroughly?
If you have got info about it, please contact me at franciscozabala@yahoo.com

Thanks ;)