Sibilants

[AnTeallach]

(Reposted from this thread.)

Because I was not sure what the dots were supposed to signify. For instance, I would not have assumed that there would have been any apical palatals like that, so that first dot must have been marking a dental POA and not an alveolar POA.

Ladefoged and Maddieson, in The Sounds of the World's Languages describe Toda as having a sub-apical palatal which actually contrasts with an apical postalveolar, so the language has two types of "retroflex" sibilant, which I suppose would correspond to Canepari's pictures B and F. (I confess to a fairly strong anti-Canepari prejudice, though the pictures seem plausible enough.)

Ladefoged and Maddieson take "palato-alveolar" to mean "apical or laminal domed post-alveolar", and "alveolo-palatal" to mean "laminal palatalised post-alveolar", with "domed" meaning some raising of the tongue behind the constriction but not as much as in a fully palatalised sound. They also talk about "laminal flat post-alveolar" sounds, in e.g. Polish (sz), Mandarin and Ubykh, which don't have the doming at all and are sometimes called "retroflex" even though they're laminal. They transcribe such sounds with an [s] with an underdot.

For what it's worth, which probably isn't very much, given that it's based on introspection, what I have in English seems to be as follows.

- /s/ is generally a laminal alveolar.
- /ʃ/ is generally an apical domed postalveolar. Not dissimilar to picture E, but I think that it's more apical than that.
- The affricate /tʃ/ is generally a laminal postalveolar, also domed. (I don't think I can feel what my tongue is doing well enough to distinguish between pictures A and C.)
- Historic /tr/ starts with an affricate which is also a laminal postalveolar, but seems to be less domed than for typical /tʃ/, suggesting something like the laminal flat post-alveolars mentioned above. I still tend to identify it with /tʃ/, though. /s/ assimilates to the same type of articulation in words like "strange".
- Historic /tj/ (as in "tune") is some sort of apical postalveolar affricate. Again /s/ assimilates to this type of articulation in words like "stew", "stupid". Comparing the beginnings of "stew" and "shoe" suggests that my /ʃ/ articulation is more domed than my /tj/ one, which seems slightly surprising.

(Plus voiced/lenis counterparts, of course.)

[AnTeallach]

... and for a bit of context here are the Canepari pictures referred to, again from that thread:

Travis, let's carry out an experiment. The image below shows six sibilant fricatives. First, tell us which ones -- whether with a secondary articulation or not -- do you have in your idiolect and in what context. Second, assign to all of them names and symbols you use. Be as accurate as you desire.

[finlay]

When I try to pronounce the difference between B and F, it's very difficult for me not to get something whistled for F. This may be because one of my teeth isn't in line with the others, or something, but it's worth noting, perhaps?

[ol bofosh]

Are there any more illustrations like that?

[finlay]

Ladefoged and Maddieson have a few, IIRC.

I should also point out that they're somewhat of a simplification, because they don't have a third dimension – the fricative is also affected by the head-on cross section of your tongue, which has a narrow groove for sibilants and a wide slit for non-sibilants, or something like that. (that's why you can get dental s and alveolar θ) Presumably you first assume that they have the cross-section relevant for sibilants, or whatever...

[Bristel]

Ok, first of all, why is there anti-Canepari prejudice in the first place? I don't understand.

[Travis B.]

And to copy on my part...

Because I was not sure what the dots were supposed to signify. For instance, I would not have assumed that there would have been any apical palatals like that, so that first dot must have been marking a dental POA and not an alveolar POA.

Ladefoged and Maddieson, in The Sounds of the World's Languages describe Toda as having a sub-apical palatal which actually contrasts with an apical postalveolar, so the language has two types of "retroflex" sibilant, which I suppose would correspond to Canepari's pictures B and F. (I confess to a fairly strong anti-Canepari prejudice, though the pictures seem plausible enough.)

Somehow I am not too surprised this exists somewhere, even though I was not expecting Przemyslaw to bring it up as an example.

Ladefoged and Maddieson take "palato-alveolar" to mean "apical or laminal domed post-alveolar", and "alveolo-palatal" to mean "laminal palatalised post-alveolar", with "domed" meaning some raising of the tongue behind the constriction but not as much as in a fully palatalised sound. They also talk about "laminal flat post-alveolar" sounds, in e.g. Polish (sz), Mandarin and Ubykh, which don't have the doming at all and are sometimes called "retroflex" even though they're laminal. They transcribe such sounds with an [s] with an underdot.

These essentially are as they I understand them to be, even though I am not all that familiar with laminal flat postalveolars (even though I have seen them marked as retroflexes).

For what it's worth, which probably isn't very much, given that it's based on introspection, what I have in English seems to be as follows.

- /s/ is generally a laminal alveolar.
- /ʃ/ is generally an apical domed postalveolar. Not dissimilar to picture E, but I think that it's more apical than that.
- The affricate /tʃ/ is generally a laminal postalveolar, also domed. (I don't think I can feel what my tongue is doing well enough to distinguish between pictures A and C.)
- Historic /tr/ starts with an affricate which is also a laminal postalveolar, but seems to be less domed than for typical /tʃ/, suggesting something like the laminal flat post-alveolars mentioned above. I still tend to identify it with /tʃ/, though. /s/ assimilates to the same type of articulation in words like "strange".
- Historic /tj/ (as in "tune") is some sort of apical postalveolar affricate. Again /s/ assimilates to this type of articulation in words like "stew", "stupid". Comparing the beginnings of "stew" and "shoe" suggests that my /ʃ/ articulation is more domed than my /tj/ one, which seems slightly surprising.

(Plus voiced/lenis counterparts, of course.)

It is interesting that your default /s/ sound is laminal rather than apical*, whereas that I am used to is clearly apical* and actually in places contrasts with its laminal counterpart, i.e. non-initial /ss/ is the apical* [s̺ː] and non-initial /st/ is frequently the laminal [s̻ː], while initial /s/ is the apical* [s̺] and in some idiolects, e.g. my sister's and sometimes my ex's, initial /st/ is the laminal [s̻].

For the sake of completeness, though, I'll match this with:

- /s/ is either (by default) an apical* or a laminal alveolar, as determined by allophony
- /ʃ/ is either (by default) a palatoalveolar or an alveolopalatal, as determined by allophony
- The affricate /tʃ/ is either (by default) a palatoalveolar or an alveolopalatal, as determined by allophony, with the stopping POA being postalveolar
- Historical /tr/ is either a palatoalveolar affricate followed by a postalveolar/velar approximant with epiglottalization or an apical undomed postalveolar affricate followed by the same, depending on the speaker; I have the former, while the latter is more conservative
- Historical /tj/ is not handled as a special case except the /t/ may be turned into a palatoalveolar or alveolopalatal affricate, where then the allophony between the two applies, if the /j/ had not already been lost, or the whole /tj/ may be turned into one, where then the allophony with alveolopalatals then also applies

* Now that I think of this, this might be often not quite apical apical but some kind of sublaminal, where the blade is involved, but the blade is towards the tip, contrasting with, well, laminal laminal, which is further back on the blade and actually typically involves a degree of palatalization, to the point that I used to actually mark this distinction as unpalatalized versus palatalized.

[Travis B.]

Ok, first of all, why is there anti-Canepari prejudice in the first place? I don't understand.

It is because he has his own parochial ultra-narrow-distinction variant upon IPA called canIPA that no one else uses and which describes distinctions no one else makes, to a degree that makes things like my own transcription practices seem extremely broad in comparison.

[Bristel]

Ok, first of all, why is there anti-Canepari prejudice in the first place? I don't understand.

It is because he has his own parochial ultra-narrow-distinction variant upon IPA called canIPA that no one else uses and which describes distinctions no one else makes, to a degree that makes things like my own transcription practices seem extremely broad in comparison.

Ah, I remember seeing that. Yes. It is rather unnecessary.

[Travis B.]

What I wonder about is just how widespread this apical/laminal alveolar and palatoalveolar/alveolopalatal postalveolar alternation is in English, and just how does this vary over various areas. For instance, this seems to be found in some sort in most GA-ish varieties, from listening to them.

However, things do vary; for instance, I have heard both people speaking GA-ish varieties and people speaking some sort of Inland North/North Central-ish dialect from Wisconsin, who defaulted to the laminal alveolars and alveolopalatal postalveolars in codas*. Likewise, all the laminal alveolars and alveolopalatal postalveolars found in clusters, as found in my dialect, might be a more dialectal feature than is their being found before /uː ʊ w/, which appears to be extremely widespread, and is their being found before /ər/, which also appears to be common.

* This is quite jarring to hear in what is otherwise GA, much more so than in Inland North/North Central-ish dialects...

[AnTeallach]

And to copy on my part...
It is interesting that your default /s/ sound is laminal rather than apical*, whereas that I am used to is clearly apical* and actually in places contrasts with its laminal counterpart, i.e. non-initial /ss/ is the apical* [s̺ː] and non-initial /st/ is frequently the laminal [s̻ː], while initial /s/ is the apical* [s̺] and in some idiolects, e.g. my sister's and sometimes my ex's, initial /st/ is the laminal [s̻].

Ladefoged and Maddieson say that English /s/ can be either. They quote a study of British English where seven out of eight had a laminal /s/, and a study of American English (Dart 1991) where roughly half the tokens were laminal and half apical. They don't give any indication of any allophonic variation, but then the book isn't principally about English so I wouldn't expect them to go into detail.

Some other languages are more consistent; e.g. Toda /s/ is always laminal. (Toda features quite a bit in that section.)

[finlay]

And to copy on my part...
It is interesting that your default /s/ sound is laminal rather than apical*, whereas that I am used to is clearly apical* and actually in places contrasts with its laminal counterpart, i.e. non-initial /ss/ is the apical* [s̺ː] and non-initial /st/ is frequently the laminal [s̻ː], while initial /s/ is the apical* [s̺] and in some idiolects, e.g. my sister's and sometimes my ex's, initial /st/ is the laminal [s̻].

Ladefoged and Maddieson say that English /s/ can be either. They quote a study of British English where seven out of eight had a laminal /s/, and a study of American English (Dart 1991) where roughly half the tokens were laminal and half apical. They don't give any indication of any allophonic variation, but then the book isn't principally about English so I wouldn't expect them to go into detail.

Some other languages are more consistent; e.g. Toda /s/ is always laminal. (Toda features quite a bit in that section.)

I definitely have a laminal s; I find it hard to pronounce an apico-alveolar s, possibly because one of my teeth is slightly out of line, as I mentioned before.

The other one that surprised me from L&M was that apparently most BrE speakers have a post-dental fricative and most Californians have an inter-dental fricative for /θ/ (I have an inter-dental).

[AnTeallach]

The other one that surprised me from L&M was that apparently most BrE speakers have a post-dental fricative and most Californians have an inter-dental fricative for /θ/ (I have an inter-dental).

As far as I can tell, mine is apical (or possibly laminal but very close to apical) dental.

[Travis B.]

And to copy on my part...
It is interesting that your default /s/ sound is laminal rather than apical*, whereas that I am used to is clearly apical* and actually in places contrasts with its laminal counterpart, i.e. non-initial /ss/ is the apical* [s̺ː] and non-initial /st/ is frequently the laminal [s̻ː], while initial /s/ is the apical* [s̺] and in some idiolects, e.g. my sister's and sometimes my ex's, initial /st/ is the laminal [s̻].

Ladefoged and Maddieson say that English /s/ can be either. They quote a study of British English where seven out of eight had a laminal /s/, and a study of American English (Dart 1991) where roughly half the tokens were laminal and half apical. They don't give any indication of any allophonic variation, but then the book isn't principally about English so I wouldn't expect them to go into detail.

I am frankly surprised that no one we are aware of has actually stumbled upon said apical/laminal alveolar (and palatoalveolar/alveolopalatal) allophonic alternation before in North American English, though; for some reason I feel like someone somehow must have encountered this before and recorded it, and would be very surprised if I actually "found it"...

[AnTeallach]

I am frankly surprised that no one we are aware of has actually stumbled upon said apical/laminal alveolar (and palatoalveolar/alveolopalatal) allophonic alternation before in North American English, though; for some reason I feel like someone somehow must have encountered this before and recorded it, and would be very surprised if I actually "found it"...

So, are you sure that that's what's going on? Are you sure that what you're hearing from other people isn't some acoustic effect caused by the conditioning sounds? There are plenty of other North American English speakers on here; do any of them recognise this apical/laminal alternation?

By the way, though sort of on the same topic, does anyone know of a website with recordings of the distinction between apical and laminal sibilants in Basque?

[tezcatlip0ca]

Ok, first of all, why is there anti-Canepari prejudice in the first place? I don't understand.

It is because he has his own parochial ultra-narrow-distinction variant upon IPA called canIPA that no one else uses and which describes distinctions no one else makes, to a degree that makes things like my own transcription practices seem extremely broad in comparison.

It's not that narrow (and I'm never putting up my NPS-IPA extension of it on the internet, which itself is about 100 times broader than the ad-hoc thing I had at age 6, before I discovered IPA), it's slightly broader than Travis, but it is true that his statements are the work of a crackpot (read this, for example).

[Travis B.]

I am frankly surprised that no one we are aware of has actually stumbled upon said apical/laminal alveolar (and palatoalveolar/alveolopalatal) allophonic alternation before in North American English, though; for some reason I feel like someone somehow must have encountered this before and recorded it, and would be very surprised if I actually "found it"...

So, are you sure that that's what's going on? Are you sure that what you're hearing from other people isn't some acoustic effect caused by the conditioning sounds? There are plenty of other North American English speakers on here; do any of them recognise this apical/laminal alternation?

To me at least, in many cases it seems to have a very clear articulatory basis; personally this is actually more apparent than the actual acoustic difference.

For instance, following /uː ʊ w/ very clearly raise the tongue behind all coronal POAs, as does /ər/ in at least some dialects such as my own (where it is [ʁ̩ˤ] there). These clearly have a direct effect on the articulation of the preceding coronals.

What is more problematic here is that my dialect also has that constrast in many consonant clusters some of which this same means cannot be at work; it should also be noted that this is much more clearly dialectal than the above. The most obvious cases are ones of /k/, /ɡ/, /ŋ/, /r/, or /l/ followed by any coronal, usually a sibilant but sometimes also /t/, /d/, or /n/, where the tongue essentially tries to bridge from the dorsal POA with a raised tongue, resulting an laminal alveolar or alveolopalatal POA. But there are other less obvious cases that are extremely common, and some strange exceptions that make it hard to explain.

The most common other cases, where /S/ is any or /s/, /z/, /ʃ/, /ʒ/, are /Sp/~/Sb/, /St/~/Sd/, /Stʃ/~/Sdʒ/, /Sf/~/Sv/, /Sθ/~/Sð/, /Sm/, /Sn/, /Sl/, /Sr/, /Sw/, /Sj/, where the sibilant is laminal alveolar or alveolopalatal as relevant, and if also applicable, any other alveolars or postalveolars are too. (It should be noted that this assimilates through any adjacent alveolars and postalveolars.) For these the most plausible simple answer is that in most clusters laminals are simply "easier" to articulate when combined with something else, which is what it feels like subjectively to me, even though I do not know if this is actually true.

Yet there is the problem of the exception of /Sk/~/Sɡ/, where a clearly apical alveolar and palatoalveolar postalveolar articulations are favored for the sibilant. The only articulatory answer I can think of is that this makes it easier to avoid palatalizing the /k/~/ɡ/. For some reason this it is hard not to think of High German Sklaverei, pronounced with an /s/ rather than /ʃ/, here, which would imply that these consonant clusters are actually due to (definitely real) substratum influence, but I really have no actual reason to believe this.

However, this cannot just a matter of the mechanics of articulation, though, as because people commonly assimilate /st/~/sd/ clusters non-initially into [s̻ː], contrasting with /ss/ in these positions, which is [s̺ː], and some people also do this initially as well, turning /st/ into [s̻] and contrasting it with /s/ as [s̺] in that same position. So there must be something going on deeper into my dialect's phonology here.

By the way, though sort of on the same topic, does anyone know of a website with recordings of the distinction between apical and laminal sibilants in Basque?

Not offhand.

[AnTeallach]

I am frankly surprised that no one we are aware of has actually stumbled upon said apical/laminal alveolar (and palatoalveolar/alveolopalatal) allophonic alternation before in North American English, though; for some reason I feel like someone somehow must have encountered this before and recorded it, and would be very surprised if I actually "found it"...

So, are you sure that that's what's going on? Are you sure that what you're hearing from other people isn't some acoustic effect caused by the conditioning sounds? There are plenty of other North American English speakers on here; do any of them recognise this apical/laminal alternation?

To me at least, in many cases it seems to have a very clear articulatory basis; personally this is actually more apparent than the actual acoustic difference.

For instance, following /uː ʊ w/ very clearly raise the tongue behind all coronal POAs, as does /ər/ in at least some dialects such as my own (where it is [ʁ̩ˤ] there). These clearly have a direct effect on the articulation of the preceding coronals.

OK, now I'm a bit confused. I don't disbelieve you as far as your idiolect is concerned, but how can you tell that other North American English speakers are doing the same thing (which is what I thought you were claiming)?

[Travis B.]

I am frankly surprised that no one we are aware of has actually stumbled upon said apical/laminal alveolar (and palatoalveolar/alveolopalatal) allophonic alternation before in North American English, though; for some reason I feel like someone somehow must have encountered this before and recorded it, and would be very surprised if I actually "found it"...

So, are you sure that that's what's going on? Are you sure that what you're hearing from other people isn't some acoustic effect caused by the conditioning sounds? There are plenty of other North American English speakers on here; do any of them recognise this apical/laminal alternation?

To me at least, in many cases it seems to have a very clear articulatory basis; personally this is actually more apparent than the actual acoustic difference.

For instance, following /uː ʊ w/ very clearly raise the tongue behind all coronal POAs, as does /ər/ in at least some dialects such as my own (where it is [ʁ̩ˤ] there). These clearly have a direct effect on the articulation of the preceding coronals.

OK, now I'm a bit confused. I don't disbelieve you as far as your idiolect is concerned, but how can you tell that other North American English speakers are doing the same thing (which is what I thought you were claiming)?

Basically because I can not only hear them, but knowing what someone will say, can reliably pick which someone will have, for these particularly before /uː ʊ w/. This is unlike in the many clusters I speak of, where this is not true, which makes me suspect that this in clusters is more dialectal in nature..

[AnTeallach]

Basically because I can not only hear them, but knowing what someone will say, can reliably pick which someone will have, for these particularly before /uː ʊ w/. This is unlike in the many clusters I speak of, where this is not true, which makes me suspect that this in clusters is more dialectal in nature..

I feel like I'm going round in circles here...

Why are you so sure that what you're hearing is an apical/laminal variation (a pretty subtle distinction, unless you're Basque) rather than anticipatory coarticulation with the conditioning sounds (which, from what you're saying, are all velar or similar, assuming "bunched" r)?

[Travis B.]

Basically because I can not only hear them, but knowing what someone will say, can reliably pick which someone will have, for these particularly before /uː ʊ w/. This is unlike in the many clusters I speak of, where this is not true, which makes me suspect that this in clusters is more dialectal in nature..

I feel like I'm going round in circles here...

Why are you so sure that what you're hearing is an apical/laminal variation (a pretty subtle distinction, unless you're Basque) rather than anticipatory coarticulation with the conditioning sounds (which, from what you're saying, are all velar or similar, assuming "bunched" r)?

The reason why I believe this variation is so is the aforementioned medial/final /ss/ versus /st/ contrast and, for some (for all excluding full assimilation), initial /s/ versus /st/ contrast in my own dialect, which are articulated the same way, yet which clearly are not velar anticipatory coarticulation, but rather assimilation together of two phonemes both sharing the same POA without any coarticulations of their own.

Not that these are not anticipatory coarticulation - of course they are - it is just that they are manifested by the same means of apical/laminal alveolar and palatoalveolar/alveolopalatal postalveolar articulations.

[Travis B.]

It should also be emphasized that the anticipatory coarticulation almost came first here, considering that it seems quite widespread, unlike the forms found in consonant clusters which, if coarticulation, is only obvious in the cases of coronals preceded by /k/, /ɡ/, /ŋ/, /r/, or /l/, and which is much more highly dialectal in nature, and has the oddity of not coarticulating notably in the case of /Sk/~/Sg/, where /S/ is a sibilant.