Talk:Affricate consonant

old discussions

How can you tell the difference between Polish /tS/ and /t+S/ clusters?

I'm a linguist, not a native Pole, but the way I pronounce them trz [t-S] takes longer than cz [tS], and in fact some people analyse trz as [tSS], i.e. affricate [tS] followed by [S]. Another possible distinction (not being Polish, I can't tell for sure which is true) is that Polish [t] is not homorganic with [S], so there's a perceptible shift in articulation as the tongue moves back to [S]; whereas the affricate conventionally written [tS] does not really begin with a [t] segment but with a postalveolar [c+]. Gritchka

I suppose the distinction is rather whether the stop element has a release on its own or whether the release is simultaneous with the friction which makes it an affricate. I don't know Polish either, but this idea of mine has been accepted by a Ukrainian linguist.

What your Ukrainian linguist is describing is a syllabic boundary between the plosive and the sibilant. This occurs in English too, at least with word boundaries (why choose and white shoes, for example, but this isn't a good example because their are different [aj] phonemes in the two words). The Polish difference in spelling is etymological. Once upon a time, Polish rz was similar to Czech ř. Often people "hear" orthographic differences that aren't realized in pronunciation. (Many Japanese imagine they can hear a difference between oo and ou, for example, though there's none there.) I'd like to see data that Poles can reliably distinguish minimal pairs, and that the fricative parts are actually the same. However, I wouldn't be surprised if Klallam really does make this distinction, not with it being so close to an area where sequences like [pktts] and [sxs] are unremarkable words! kwami 08:14, 2005 July 11 (UTC)

Sure it is the same phoneme, and please, if you’re dealing with phonemes, make sure to put them in slashes! Flofl. 08:59, 2 September 2005 (UTC)

Once upon a time, Polish rz was similar to Czech ř

None of these two sounds is or ever was an affricate. Czech ř was originally also spelled rz - later a small z was written over rather than next to the letter, and even later it changed into haček in č ř š.... In the same time Polish retained the old spelling cz rz sz. What it is supposed to have to do with the topic?

One of the sequences discussed was trz, and this was being contrasted with the others with the claim that the difference was stop+fricative vs affricate. My point was that if the fricatives rz and sz were different (other than voicing), that would explain the minimal pairs. (It appears now that they aren't different, but that was my question.)

I'd like to see data that Poles can reliably distinguish minimal pairs, and that the fricative parts are actually the same.

But minimal pairs are clearly distinguishable. Let's cite Swan :

Contrast trzy “tszy” or “czszy” to czy “czy”. Pronunciations such as “czeba” for trzeba occur, but are considered careless and substandard (pronounce instead “tszeba” or “czszeba”).

Good. That's a nice reference.

Actually, he contrasts *three* phoneticaly different possibilities here, "tszy", "czszy" and "czy" ([tʃɨ], [t͡ʃʃɨ] and [t͡ʃɨ], or in another transcription tšy, čšy and čy)...

But not phonemically distinct. It sounds like there's a range of assimilation for the /t/, so it wouldn't be possible to have a minimal triplet. kwami 21:12, 2005 August 2 (UTC)

I put up the Klallam info. It is from Timothy Montler (http://www.ling.unt.edu/~montler/), who does fieldwork on the language. He has a textbook for high school students currently in progress. The [t] + [s] sequence results from a /-t/ subject-in-control suffix (in transitive verbs) followed by a /-s/ 3rd person subject suffix (number unspecified). The affricate is a fused morpheme that is a combination of the control suffix and a 1st/2nd person object suffix. This contrast exists for all transitive verbs. All objects are suffixes, all other subjects are post-predicate enclitics, 3rd person subject (-s) is a suffix. Thus, we have

• sáʔəts "she lifts it",
• sáʔət st "we lift it",
• sáʔəc st "we lift you",
• sáʔəc cxʷ "you lift me",
• sáʔət cxʷ "you lift it"

Note the nice consonants clusters: [tst], [t͡sst], [t͡st͡sxʷ], [tt͡sxʷ]!

In Klallam a sequence of [t]+[t͡s] or [t]+[t] is pronounced with the first [t] being released followed by the [t͡s] or [t], and not as [tː͡s] or [tː], i.e. with an initial closure of a longer duration (unlike many other languages, such as Japanese). You can see this in ʔítt "he sleeps".

The same contrast is also present in definite articles/demonstratives: cə (IPA: [t͡sə]) "specific, visible" (gender unmarked); tsə (IPA: [tsə]) "specific, visible, feminine". So, cə cačc "uncle or aunt", tsə cačc "aunt".

All non-glottalized stops & affricates are aspirated (much like English). (Montler's site has sound files.) So, perhaps one could transcribe the difference between cə & tsə as [tʰsə] & [tsə]. But, aspiration (or voicing) is not contrastive anywhere else. Describing the difference as a matter of release makes more sense. This is the analysis of Montler. (by the way, the closely related Saanich dialect of North Straits Salish has unaspirated or barely aspirated stops & affricates. Montler says that it is noticebly different from Klallam in this respect.)

Another language on the Northwest Coast, Oowekyala (Wakashan), has this contrast. See Darin Howe's (2000) dissertation (http://depts.washington.edu/wll2/files/howe_00_diss.pdf) for examples (or maybe I will put them here). Stops are aspirated in this language as well.

– ishwar  (speak) 15:16, 2005 July 11 (UTC)

I think that the separate release is the more marked case. So normally, when writing [ts] we read it as if there were no audible release of the [t] (the same as e.g. [nd], that is normally read as if there were no audible release of the [n]). Unfortunately, the IPA has no way to explicitly mark the audible release, but only a way of marking the lack of it. J. 'mach' wust 01:00, 17 Nov 2004 (UTC)

Actually, it does! The ExtIPA has a symbol, t⁼ (here on a tee), that indicates non-aspirated release. However, if Klallam is like other languages in the Pac NW, it is probly aspirated. kwami 08:14, 2005 July 11 (UTC)

According to IPA in Unicode, the sign ⁼ marks unaspirated consonants in extended IPA transcription of disordered speech. However, I don't know what extended IPA is. -- j. 'mach' wust ^{ˈtʰɔ̝ːk͡x} 10:37, 11 July 2005 (UTC)

The extIPA is a group of symbols that are were developed by and are primarily used by speech language pathologists for transcribing disordered speech. I havent seen these used in much linguistic work. peace – ishwar  (speak) 15:32, 2005 July 11 (UTC)

No, not much used (or they'd be in the regular IPA). But they're there if you need them. Probably be a good idea to include an explanatory note, however. kwami 01:44, 2005 July 12 (UTC)

Are there sound files for these? In the article, what're the f. and vertical line? lysdexia 15:16, 29 Nov 2004 (UTC)

The f. must be the abbreviation for femininum, that is for the grammatical gender. And the vertical line separates segments. :-)

That use of the vertical line is not described in the IPA-chart. According to the IPA chart, it's the affricate that should be marked with a tie bar.

Oh, I wanted just to write a little reply, but now I've made a revision of the whole article... Hope it answers your questions as good as possible without sound files. J. 'mach' wust 23:11, 29 Nov 2004 (UTC)

See comments above. (I really need to look into this myself.) kwami 08:14, 2005 July 11 (UTC)

palatalized affricates

Hi Vuo,

I'd left out 2ary articulations. Lots of languages have labialized, pharyngealized, etc. affricates. However, a "palatalized [tʃ]" in the IPA sense means a [tɕ], which is already covered. (Palatalized [ts] of course is another story) I'd prefer to remove your comment unless we have a clear contrast between [tʃʲ] and [tɕ]; even then it's likely that we'd have a phonemic [tʃ]-[j] cluster, and not coarticulated palatalization. Can you come up with an example of palatalized [ts] instead? Then I've got some interesting labialized affricates to add. kwami 19:21, 2005 August 12 (UTC)

Actually, the source (fi.wikipedia) indeed calls the 'palatalized' tsh an alveolopalatal affricate, so nothing unusual there. I'm not really sure if there is any frication at the channel between the palatal vault and the tongue, though. It's a modification of pitch, mainly. --Vuo 22:47, 14 August 2005 (UTC)

No, there wouldn't be a second source of frication. That's never been demonstrated for any language (though it's been claimed for a few). kwami 00:05, 2005 August 15 (UTC)

Good, as this removes one possible source of confusion. Such a fricative, if existent, would be highly unusual for a Baltic-Finnic language. But, as for palatalized 'ts', most Uralic languages recognize it, if they recognize 'ts'. Generally in Uralic, /i/ and /j/ effect palatalization on a neighboring affricate, but sometimes the cause is deleted and effect remains, retaining the phonemic status that the deleted (semi)vowel held. --Vuo 22:27, 15 August 2005 (UTC)

Yes, I believe that [tsʲ] is common both in Fennic and in the Balto-Slavic languages they influenced. What's unusual in this part of the world is that these have remained alveolar for so long. kwami 23:23, 2005 August 15 (UTC)

Postalveolars are distinct sounds, recognized by at least the Finnic proto-language. Furthermore, they do not have a higher pitch and a more "childish" sound, which is the cue for recognizing palatalization; quite the contrary, they have a lower, "angry" sound. They appear more velarized than palatalized, actually. And, in standard Finnish (but not deep Eastern Karelian), postalveolars are seen as unusual or disordered speech (an effect of drunkenness, speech defect, foreign influence or such). --Vuo 11:19, 16 August 2005 (UTC)

Alveovelar

[Because of a note in the history] Just to say that even if alveovelar is not (yet) a real world, alveovelar phones sure do exist!! Good phoneticians in several different languages did create the term, as it is very useful. Flofl. 08:48, 2 September 2005 (UTC)

Well, if it is a real word, neither I nor Google have ever heard of it. But if there are such things, [t] and [d] are not it. I look forward to reading more about them. Nohat 09:06, 2 September 2005 (UTC)

What do you call the [ɫ] phone of portuguese? Alveovelar refers to a contoid articulated simultaneously in the alveolar and velar places. Similarly, the official IPA has [w] for the labial-velar approximant. Flofl. 11:34, 5 September 2005 (UTC)

I think the usual term for [ɫ] is velarized alveolar. Nohat 18:46, 5 September 2005 (UTC)

That was my fault. Sorry to send you off on a wild goose chase, Nohat. It was simply a typo for alveolar that I didn't catch.

As for a true alveo-velar sound, how about [k!]? kwami 09:31, 2005 September 2 (UTC)

Postalveolar vs. palato-alveolar; uvular

First, I removed:

• [[voiceless postalveolar affricate|voiceless palato-alveolar affricate]]
• [[voiced postalveolar affricate|voiceless palato-alveolar affricate]]

pointing out that "postalveolar" and "palato-alveolar" are usually held to be synonyms (for example, palato-alveolar consonant redirects to postalveolar consonant, and the two terms are used synonymously on that page). Kwamikagami changed it to:

• [[voiceless retroflex affricate|voiceless postalveolar affricate]] [t̠s̠]
• [[voiced retroflex affricate|voiceless postalveolar affricate]] [d̠z̠]
• [[voiceless postalveolar affricate|voiceless palato-alveolar affricate]] [t̠ʃ]
• [[voiced postalveolar affricate|voiced palato-alveolar affricate]] [d̠ʒ]

which is just confusing because the reader sees "postalveolar" but if he clicks on it he gets led to the page for "retroflex", not to the page for "postalveolar" (for which he would have to click on "palato-alveolar"). And anyway, [t̠s̠] and [d̠z̠] aren't retroflex either. The diacritic under them is the diacritic for retraction, so these symbols would have to stand for sounds made immediately behind the alveolar ridge; I doubt they are distinct from the alveolo-palatal affricates [t̠ɕ], [d̠ʑ]. I recommend we not have any piped links right here, and just have:

• [[voiceless alveolo-palatal affricate]] [t̠ɕ]
• [[voiced alveolo-palatal affricate]] [d̠ʑ]
• [[voiceless postalveolar affricate]] [t̠ʃ]
• [[voiced postalveolar affricate]] [d̠ʒ]
• [[voiceless retroflex affricate]] [ʈʂ]
• [[voiced retroflex affricate]] [ɖʐ]

Second, I added:

• voiced uvular affricate [ɢʁ] (needs confirmation)

noting that it is probably a hypothetical sound. Kwamikagami removed it, saying "Why add a hypothetical sound?" A fair enough question, but I don't think [ɢʁ] is any more hypothetical than [ɟʝ] or [ɡɣ], which are also already listed. I say, either all three hypothetical voiced dorsal affricates should be listed, or none. What say ye? --Angr/_{tɔk tə mi} 14:03, 22 September 2005 (UTC)

Okay, in reverse order: I didn't remove [ɟʝ] or [ɡɣ] because I thought the person who originally added them might have known what they were doing. I merely tagged them as needing to be confirmed. But we're moving the wrong way if we intentionally add unattested sounds.

[t̠s̠] is frequently called 'retroflex', as in Polish and Mandarin, and this will be explained if the article is ever written. [t̠ʃ] is not a simple postalveolar, it is in addition semi-palatalized. ([t̠ɕ] is fully palatalized, and contrasts phonemically with [t̠s̠] in both Polish and Mandarin.) All non-palatalized postalveolars are called 'retroflex' in at least some of the literature. [t̠s̠] and [ʈʂ] can be synonymous, but they are sometimes specifically used for laminal and apical postalveolars. (Pre-palatal [ʈʂ] is sub-apical.) kwami 14:19, 22 September 2005 (UTC)

P.S. Personally, I would like to restrict the postalveolar articles to [s̠] and its relatives, and create separate palato-alveolar articles for [ʃ] et al, but I didn't think people would go for it. In any case, purely postalveolar [s̠] can be covered under 'retroflex'. kwami

As I understand it, palatalization means the raising of the blade of the tongue in the direction of the hard palate while some other part of the vocal tract is making the primary articulation. Thus any sound like [(t̠)ʃ] and [(t̠)ɕ] made with the blade of the tongue as the primary articulation cannot be palatalized as a secondary articulation. Also, as I understand it, although it is true that Polish and Mandarin retroflexes are articulated rather differently from Indo-Aryan and Dravidian retroflexes (and for that matter rather differently from each other), the various types of retroflex never contrast with each other within a single language. Therefore there is no reason not to use the symbol [(ʈ)ʂ] for them all. The symbol [(t̠)s̠] is just ambiguous, since the retraction symbol says nothing about apicality, laminality, or subapicality. (That's why I misinterpreted how you were using it in the first place!)

To make a long story short, at the moment, we have the following articles about voiceless sibilant fricatives (as well as articles for their voiced counterparts):

• Voiceless alveolar fricative - s
• Voiceless alveolo-palatal fricative - ɕ
• Voiceless postalveolar fricative - ʃ
• Voiceless retroflex fricative - ʂ

In other words, articles for each of the sounds for which the IPA provides a distinct diacriticless symbol. I would like this page to list all and only the affricate correspondents of those fricatives, namely:

• Voiceless alveolar affricate - ts
• Voiceless alveolo-palatal affricate - t̠ɕ
• Voiceless postalveolar affricate - t̠ʃ
• Voiceless retroflex affricate - ʈʂ

and of course their voiced counterparts. --Angr/_{tɔk tə mi} 15:27, 22 September 2005 (UTC)

Palatalization is the exact opposite; it means that in addition to a primary articulation, the center of the tongue is raised, which makes the channel narrower and the pitch higher. For example, Russian contrasts the postalveolar fricatives Ш (plain) and Щ (palatalized). Unfortunately, the term "palatalization" is used also by linguists other, conflicting meanings. --Vuo 18:28, 22 September 2005 (UTC)

Okay, I removed the retracted alveolars. Also extended the definition of the retroflex consonants to cover Slavic and Chinese. kwami 23:43, 22 September 2005 (UTC)

Trilled affricates

Are "trilled affricates" really affricates? I've only seen them referred to as "trill-released stops". --Ptcamn 23:01, 5 June 2006 (UTC)

Palatal stops

Within the IPA, [tʃ] and [dʒ] are sometimes transcribed as palatal stops, <c> and <ɟ>.

What? Surely postalveolar affricates and palatal stops are different enough to warrant separate notation. (E.g., they're different phonemes in Hungarian.) —Naddy 14:13, 31 October 2006 (UTC)

True palatal stops are rather prone to decaying into postalveolar affricates (see palatalization). They may regardless continue to phonologically behave as palatal stops (AIUI, they usually do, at least as long as there are no newer palatals getting in the way). Plus, a single symbol can be more convenient to use… --Trɔpʏliʊm • blah 14:07, 27 August 2008 (UTC)

Swiss-German qχ?

Does anyone know anything about this?Cameron Nedland 18:22, 6 July 2007 (UTC)

I'm not personally able to confirm its existence (I can't remember hearing a Swiss affricate that immediately struck me as obviously uvular), but it appears plausible that it does exist as a dialectal or idiolectal allophone of the velar affricate. Searching for "velar affricate", I've found a posting on Linguist List which shows that a lot of linguists find this interpretation credible, and even likely. --Florian Blaschke (talk) 19:47, 17 June 2011 (UTC)

Heteroorganic affricates

This topic could use a whole subsection. Mainly, on when they exist and when they don't. My understanding is that they by default aren't, unless there are specific reasons to consider them as such - so eg. tax and depth would not be applicable for affricate status - but I've never found a good treatise of the subject. Still, it keeps irking me whenever I hear the letter X described as an "affricate".--Trɔpʏliʊm • blah 14:24, 27 August 2008 (UTC)

The palatal sound typical of the Indic languages such as Hindi, usually transliterated "c", has been described as phonetically a hetero-organic affricate [cʃ] e.g. by Masica, Indo-Aryan Languages, p. 94, Google Books Link. Grover cleveland (talk) 02:38, 5 February 2009 (UTC)

although Masica is not a phonetician IIRC. Maybe he wanted to be more precise on the stop part, but did not bother that much for the fricative part. One can find these affricates transribed as t̠͡ɕ or d͡ʑ, (sometimes also with curly tail on t and d, but I cannot find them now) but the curly tails on c and z are a bit difficult to produce. My guess is that the representation as ʃ is due to convenience, and does not imply that the poa is necessarily postalveolar. Jasy jatere (talk) 09:55, 5 February 2009 (UTC)

Well, he does say "a more accurate phonetic representation would be [cʃ]" (I hope this is visible in the Google books link), which seems to imply that he is specifically concerned with the phonetics here, rather than phonology or whatever. However (thinking about it further), I think he may have meant to say [cɕ]: from my experience in Hindi, the sound is created by placing the blade of the tongue flat against the hard palate and releasing with a laminal fricative in the alveolar or postalvelar region. I'm not even sure how [cʃ] would be physiologically possible. Grover cleveland (talk) 17:04, 6 February 2009 (UTC)

"Catch it"

Maybe there should be a disclaimer about "catch it"/"cat shit". I my dialect the final /t/ in "cat" is alveolar, although I think it also has a preceding glottal stop; but in any case it isn't completely reduced to a glottal stop. I think it's like that in a lot of 'lects. The vowels in "cat" and "catch" are also different, FWIW. Finally, I'm not sure of the phonology at work, but "cat shit" seems to have more of a pause between the /t/ and the /S/. 71.90.130.7 (talk) 07:50, 12 December 2008 (UTC)

cat shit/catch it was just something at the top of my head, though there must be other minimal pairs that are maybe more widespread. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 19:48, 12 December 2008 (UTC)

*tsh* = ch?

Stop-fricative sequences may also have a syllable boundary between the two segments, syllable ??? word ???

http://www.merriam-webster.com/dictionary/Affricates A stop and its immediately following release into a fricative that are considered to constitute a single phoneme (as the \t\ and \sh\ of \ch\ in choose) t+sh= ch ?

http://www.merriam-webster.com/dictionary/sweatshirt

sweatshirt

tsh is affricate or stop-fricative sequences?

[sweʔʃərt]? [swetʔʃərt]? [swet͡ʃərt̚]? —Preceding unsigned comment added by 83.145.72.70 (talk) 17:38, 19 October 2009 (UTC)

The tsh of sweatshirt constitutes a stop-fricative sequence. Phonemically, it's /ˈswɛtˌʃɜrt/. In dialects that glottalize syllable-final /t/, it would be something like [ˈswɛtˀˌʃɜɹt̚ˀ] or [ˈswɛʔˌʃɜɹʔ]. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 20:14, 19 October 2009 (UTC)

Tshiluba, [tˀˌʃ*] or [ʔˌʃ*] or [tʃ*]???

In English, there is no word-initial contrast between a stop-fricative sequence and an affricate, so tshiluba (I'm not familiar with this term) would have the affricate. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 18:19, 20 October 2009 (UTC)

adj*, for exemple: adjective ? dj is affricate or stop-fricative sequences? —Preceding unsigned comment added by 83.145.72.70 (talk) 18:41, 24 October 2009 (UTC)

It might be better to look at a word's morphology and etymology. sweatshirt as a compound of sweat and shirt (and not because of its spelling), has the cluster. Adjective has been in English so long that if it were a cluster in the beginning it has long since become an affricate. I suspect also that since English has makes little contrast between /dʒ/ and /ʒ/ that you may not find any such clusters. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 19:40, 24 October 2009 (UTC)

Honestly, could we have a more pleasant example phrase than "cat shit"?

Know of another minimal pair? --Florian Blaschke (talk) 22:18, 17 June 2011 (UTC)

How about: watch-tick vs. what shtick? --Trɔpʏliʊm • blah 23:28, 28 June 2011 (UTC)

That would be only a near-minimal pair for speakers without the Wine–whine merger. --JorisvS (talk) 09:44, 29 June 2011 (UTC)

Isn't the "t" in "cat" from "cat shit" an unreleased stop? As in, [kʰæt̚ʃɪt̚], not [kʰæʔʃɪt̚]? 99.146.122.70 (talk) 15:23, 3 August 2011 (UTC)

Sound files, please

I see 2-column lists of sounds, but no links to .ogg files. Also, shouldn't the 2 columns be reworked as a table?

Also, it would be clearer if the coronals were distinguished with the labels laminal, apical, and subapical (and possibly a fourth group ("laminal/apical"??) for those languages (e.g. English?) where the laminal/apical distinction is not made and doesn't confuse listeners, i.e., doesn't really matter. I think this would really help with "retroflex" &c. — Solo Owl (talk) 19:12, 20 January 2011 (UTC)

Sound files added. Oops, one of four table. Three to go. -DePiep (talk) 09:39, 30 June 2011 (UTC)

Bilabial-labiodental affricates in Mandarin and velar affricates in English

According to Mandarin Chinese#Initials, voiceless bilabial-labiodental affricates (both unaspirated and aspirated) also exist in regional Mandarin dialects as regular regional variants of the retroflex affricates.

I also remember reading about voiceless velar affricates found in certain (northern?) English dialects. Scouse perhaps? Check Talk:Scouse#Evolution Or Affectation?.

According to Linguist List, Sotho-Tswana also has a voiceless velar or uvular affricate, and Sesotho phonology#Consonants confirms this at least for Sotho/Sesotho. --Florian Blaschke (talk) 22:15, 17 June 2011 (UTC)

In view of wikt:pfóltu, I wonder if Assan and possibly other Yeniseian languages might have them, too. --Florian Blaschke (talk) 06:45, 24 June 2011 (UTC)