Talk:Final-obstruent devoicing

Removed Modern Greek
There are no voiced final consonants in Modern Greek (except for foreign words), and such devoicing does not happen within words, so I am removing Greek from the list.

Where does this sentence belong?
"There are however some exceptions in very few words borrowed from English."

I can't say whether or not these exceptions occur in Afrikaans or Russian, etc. But there are certainly no such exceptions in German. It should be clarified what languages know exceptions in foreign words. —The preceding unsigned comment was added by 89.52.167.158 (talk) 09:22, 13 March 2007 (UTC).


 * I second that and remove the sentence. If someone wants to reinsert it, please add some details, such as language, circumstances, examples etc. --Neg 19:53, 23 March 2007 (UTC)
 * In German, it varies a lot depending on the speaker. Younger Germans, especially those who know English well, do make an effort to suppress final devoicing when using English words like Job and Airbag in German. —Angr 21:34, 19 April 2007 (UTC)
 * Some of them might try, but the norm is certainly to pronounce job (a very common word in German news reporting when unemployment is being discussed) like chop. --Doric Loon (talk) 14:35, 7 February 2009 (UTC)
 * Those who know will do so when actually speaking English. When speaking German, at least the words that are by now full German words, will also be pronounced in a German way. "Job" (spoken: Tschopp) certainly belongs to them.--151.100.102.140 (talk) 17:03, 7 February 2014 (UTC)

Old English
Wouldn't that rather be medial voicing of fricatives? -- Trɔpʏliʊm • blah 22:54, 8 July 2010 (UTC)
 * Diachronically at least, it's both. The complementary distribution of [f] and [v] in Old English results from (1) the devoicing of Proto-Germanic *β to [f] when it was word-final in Old English, and (2) the voicing of Proto-Germanic *f to [v] when it was intervocalic in Old English. Also, note that Proto-Germanic *x (*h) did not undergo intervocalic voicing in Old English (instead, it disappeared), but Proto-Germanic *γ did undergo final devoicing to [x] (spelled h) in Old English. +Angr 10:10, 10 July 2010 (UTC)

Final obstruent voicing distinction?
The curious thing about the devoicing, especially in the case of stops, is that it is so common. I wonder if the conventional understanding of the phenomenon is actually true, and not perhaps clouded by the circumstance that it is two extremely prominent languages, namely English and French, which (seemingly) do not feature final devoicing even in stops, and that many languages do not have a voicing distinction in stops (or even in general) in the first place, or do not even allow final stops and are thus missing from the list (which has a strong European bias anyway), giving the impression that it is a restricted phenomenon, an exception rather than the rule, when it is actually not. The reason is that phonetically speaking, both examples, English and French, are not particularly clear, especially considering stops. In both cases, final devoicing was originally present as well and the emergence of final voiced stops and fricatives is a relatively recent phenomenon (especially in French), caused by the disappearance of final schwas. Still in contemporary French, precise phonetic investigation reveals that the schwa has actually not entirely disappeared: There is still a super-short schwa in the standard pronunciation, after voiced consonants, however, it is almost completely devoiced and therefore difficult to hear. It could also be described as a vocalic release after final consonants. Strictly speaking, therefore, there are still no final consonants in French, and there are still essentially only open syllables on some level that is also detectable on the phonetic level. When speakers of French pronounce foreign words with certain consonant clusters, such as, which are originally alien to French, it can be discerned that they almost seem to break up the clusters up with tiny schwas, or at least the syllabic rhythm is (impressionistically) so that the clusters sound "slower". Therefore, it is not, strictly speaking, entirely correct to describe French as a language that distinguishes voicing in final consonants, or at least there is reason for doubt. In the case of English, the disappearance of final schwas that caused final voiced stops and fricatives to develop is much older, and the phonetic difficulty is something else: Phonetic measurements reveal that it is not primarily the voicing that distinguishes pairs like bed/bet or bad/bat, but mainly the duration of the preceding vowel, of the stop itself, and other cues such as aspiration or preglottalisation in the case of the unvoiced stop. The stop is, actually, devoiced, at the very least partially.

Therefore, it would be much more interesting to list languages, apart from English and French, which distinguish voicing in either final stops or fricatives, or in both. Moreover, it should be investigated if this distinction is strictly speaking present, or whether it has to be qualified as in the case of English and French. I have a sneaking suspicion that true final voiced stops are, if not impossible, then at least very difficult to articulate, and the same may be true for fricatives. In this case, it would be natural that the neutralisation of voicing distinctions in word-final position is extremely common in the languages of the world. --Florian Blaschke (talk) 05:22, 5 July 2011 (UTC)
 * It's true that final obstruent devoicing is extremely common. Even within English, voicing distinctions at the end of a word are often lost in children's speech and in drunk speech, and even speakers of languages with no final obstruents at all (like Mandarin) often apply final devoicing to English when they're learning. Compiling a list of languages with a word-final voicing distinction in obstruents, or just stops, might be interesting, but hard to do without a lot of research. Most published research into the phenomenon discusses languages where devoicing does happen, not ones where it doesn't.
 * I'm not sure what you mean by "In both cases [French and English], final devoicing was originally present as well and the emergence of final voiced stops and fricatives is a relatively recent phenomenon". That's true in French, but English has never had final devoicing. Old English distinguished voicing for stops word-finally, and the lack of a distinction for fricatives was due to the fact that Old English didn't have voiced fricative phonemes at all, so there was no distinction to get lost word-finally.
 * In optimality theory, the idea is that there's a markedness constraint against word- or syllable-final voiced obstruents, which is ranked with respect to a faithfulness constraint expecting distinctions to be maintained. In languages with FOD, the markedness constraint is higher ranked, while in languages without it, the faithfulness constraint is higher ranked. But according to many OT-based models of language acquisition, children start out with very high-ranked markedness constraints (which is why their pronunciation is so simplified) and then learn to promote faithfulness constraints on the basis of what they hear in their language environment. This contrasts with a rule-based model of acquisition which says that children learn rules such as FOD to apply to the underlying forms they learn. If the OT model is right, it suggests that children learning English have to learn to switch FOD off, rather than that children learning German have to learn to switch it on. The fact that English-speaking children do sometimes apply FOD, while German-speaking children (as far as I know) never fail to apply it, suggests that's true. Angr (talk) 06:17, 5 July 2011 (UTC)
 * I was thinking of cases like ME dogge > ModE dog, but you are right, OE is understood to have distinguished voicing in final stops, as well (I simply forgot that), but I can't think of any minimal pairs and all voiced stops apart from d would only occur geminated, or after nasals, in word-final position, anyway. But since we cannot recover any phonetic details from OE, or probably at least not those relevant in this context, it's not really a great counterexample. The reason I discussed Modern English and French was that the phonetic investigation suggests that the real distinction is not actually voicing. On a phonological level, one might consider an analysis where French still has final schwas (in careful pronunciation they are still much more conspicuous, not to mention special registers such as used by classical singers, for example), and where the apparent distinction in English is actually secondary to something else, as I have outlined above. --Florian Blaschke (talk) 06:30, 5 July 2011 (UTC)
 * See also Talk:Korean phonology, where the general understanding that English has voiced stops at all underlyingly is criticised: apparently the alternative is that seemingly voiced stops are really unvoiced and acquire allophonic voice in voiced environments. This also implies that seemingly voiced English word-final stops are not really voiced – the apparent voiced quality must be inferred from somewhere else, such as a preceding lengthened vowel. --Florian Blaschke (talk) 02:10, 15 February 2015 (UTC)

Split list into parts.
I wonder if the "List of languages with final obstruent devoicing" should be split into parts. The list has a high proportion of Germanic and Slavic languages, and it might be more useful to have, say, four lists: Germanic; Slavic; other IE; non-IE. This won't split the list into even quarters, but I think it will make it easier to make sense of. (And it might encourage the addition of interesting examples.) —Ruakh TALK 17:02, 1 December 2011 (UTC)


 * Thank you, CodeCat! :-)  —Ruakh TALK 23:56, 7 December 2011 (UTC)

South African English
When listening to South Africans speaking English I often hear final devoicing, which along with other differences helps to make it sound characteristically Dutch-accented (or Afrikaans-accented) to me. Is it true that South African English has final devoicing? CodeCat (talk) 15:03, 23 December 2011 (UTC)
 * Some speakers might have final devoicing, but by no means all of them. Peter238 (talk) 17:01, 26 September 2014 (UTC)
 * Afrikaans-accented SAE sounds quite different from SAE that has little or no Afrikaans influence. The kind with Afrikaans influence not only usually has final devoicing, it also usually has unaspirated stops in words like pan, tan, can. But speakers without much Afrikaans influence don't have final devoicing (at least, no more so than speakers of other varieties of English) and do have aspirated stops. —Aɴɢʀ (talk) 07:19, 28 September 2014 (UTC)

Different types of final devoicing
In a discussion on Wiktionary I noticed that even when some languages have devoicing of the same types of sounds, there might still be a difference in when that devoicing is applied. In Dutch, for example, devoicing is not purely phonetic, but has a lexical/morphological element. Words that are devoiced in isolation are also devoiced when they are part of a compound: badwater "bath water" has a voiceless [t], like bad does by itself, even though the plural baden has a voiced [d]. Similarly for avondzon "evening sun" with [ts], but avonden with [d]. On the other hand, I believe that many Slavic languages don't have this phenomenon; they devoice only in isolation, but apply voicing assimilation in clusters as a form of sandhi (I know Slovene does this, at least). Is this something that should be noted in the article? CodeCat (talk) 19:20, 16 March 2014 (UTC)


 * Yes, definitely. Otherwise readers familiar with one language may assume that others are like it.  Since you have the examples, could you do the honors?  — kwami (talk) 06:40, 17 March 2014 (UTC)
 * I only really have the example I gave, and no sources... CodeCat (talk) 17:51, 17 March 2014 (UTC)


 * Good enough. If it's challenged we can see about proving it.  — kwami (talk) 20:06, 17 March 2014 (UTC)

Phonetic or phonological?
The list at the end of the article mentions Marshallese, but also notes that this language has no voicing contrast to begin with. This feels a bit like cheating to me; allophonic variation doesn't seem like "true" final devoicing. So I'm wondering whether the usual definition of FOD concerns phonetic devoicing (the actual loss of voiced articulation) or phonemic (the loss in contrast between voiced and voiceless). In the latter case, it might be interesting to consider the wider phenomenon of obstruent collapse, not just loss of distinction in voicing. I believe that many languages have deaspiration of final consonants as well, for example. CodeCat (talk) 20:30, 23 September 2014 (UTC)
 * Yeah, it's a form of neutralisation that doesn't necessarily have anything to do with voice in the phonetic sense (compare my criticism above), so the term is misleading. As for final consonants, it's not only deaspiration but can be "de-release-ation" as in no audible release as well. If, for example, English final voicing is indicated by lengthening of the preceding vowel or also absence of preglottalisation/pre-aspiration, "devoicing" is really a type of shortening of the preceding vowel or preglottalisation etc. phonetically. --Florian Blaschke (talk) 02:20, 15 February 2015 (UTC)
 * Remember how PIE seems to have final voicing? Whatever may have happened there phonetically ... --Florian Blaschke (talk) 02:25, 15 February 2015 (UTC)

English
Removed this line:
 * Relics of a final devoicing of "v" are to be found between words so related as "half" and "halves", and "knife" and "knives" distinguishing singular and plural.

This is not due to a process of a voiced phoneme losing voicing word-finally (which is what this article is about), but rather a voiceless phoneme /f/ that had the voiced allophone [v] intervocalically in Old English. — Preceding unsigned comment added by 70.181.243.30 (talk) 11:44, 30 November 2016 (UTC)
 * Which is correct for Old English, but came about by, yes, the devoicing of Proto-Germanic *b: *halbaz (contrast e.g. German halb, Swedish halv), *knībaz (contrast e.g. Swedish kniv). -- Trɔpʏliʊm • blah 23:50, 3 December 2016 (UTC)
 * Then it belongs in a section on Proto-Germanic, not English. — Preceding unsigned comment added by 70.181.243.30 (talk) 09:48, 25 February 2017 (UTC)

Armenian
The list "other languages" includes Armenian (for stops). This is not appropriate, since devoicing occurs in an unpredictible way: Some words written with a letter usually denoting a voiced stop (and I skip the discussion about the exact phonetic nature of these stops) are pronounced with a voiceless aspirated stop (including affricates!), but there are many words with a "voiced" word-final stop/affricate, such as [bad] "duck", [meg] "fog", [t'ag] - apostrophe for aspiration - "crown", as against [muk'] "dark" or [barts'] "cushion" and many more on both sides. Remarkably, the word for "crown" with its "voiced" final stop changes this stop to a voiceless aspirate in the word [t'ak'avor] "king" = "bearer of a crown". This unpredictability is clearly shown in the "Armenian orthographic-orthoepic-terminological dictionary" by Barselyan (1973) - in Armenian -, where all deviations from the expected pronunciation are indicated in square brackets. I suggest to remove Armenian from the list, since there is no regularity with final devoicing.

Michael Job [mjob@gwdg.de] — Preceding unsigned comment added by 2A02:8108:85C0:8399:8D94:5D03:7CE2:21BC (talk) 22:47, 28 June 2018 (UTC)

Devoicing in compounds - inappropriate examples
thief-thieves, bath-bathe do not illustrate compounding. Barefoot through the chollas (talk) 16:25, 14 February 2022 (UTC)

And months later, still no examples of compounds. I'll give it a little more time, then if nothing shows up to justify the section, just delete the whole thing. Barefoot through the chollas (talk) 17:12, 23 June 2022 (UTC)