Sievers' Law-like phenomenon in English

[alice]

Warning: Moderately Extreme Geekiness follows.

While watching GoT I noticed two distinct pronunciations of Arya, viz. /ar.ja/ and /a.ri.ja/ (never mind the quality of the first vowel; it's easier to type and isn't relevant here). This is a bit like Sievers' Law in certain varieties of Indo-European, where /j/ > /ij/ after heavy syllables (thus /an.ja/ remains, but /ant.ja/ > /an.ti.ja/, although it's probably unrelated.

I've noticed a lot of similar postconsonantal /j/ becoming /ij/ in various people's pronunciations, generally of foreign words but also of nativised Latinate borrowings, and after all sorts of consonants where it wouldn't be expected; for example it occurs in "union" but not in "onion". What might be the reason for it?

[brandrinn]

It seems to be mostly limited to borrowed words with {y}, like "Tokyo," and could be the result of people just not knowing when {y} is supposed to be a consonant and when a vowel.

I've never heard "union" pronounced with three distinct syllables.

[Astraios]

I hear this all the time, but I notice it most in English-speakers trying to pronounce French or Hebrew, because when it comes to C+/j/ they just can't do it.

[ol bofosh]

It seems to be mostly limited to borrowed words with [y], like "Tokyo," and could be the result of people just not knowing when [y] is supposed to be a consonant and when a vowel.

I've never heard "union" pronounced with three distinct syllables.

I say it with three syllables.

I hear this all the time, but I notice it most in English-speakers trying to pronounce French or Hebrew, because when it comes to C+/j/ they just can't do it.

I can, but I get yod convalescence (so Wiki tells me), so duty becomes jew-ty etc.

[Nortaneous]

I don't think I've ever heard "Tokyo" pronounced with two syllables.

Anyway, I wouldn't be surprised if this is a thing; Cj clusters are pretty rare except before /u/.

[Travis B.]

If anything, I am more used to the pronunciation of Tokyo as /ˈtokijo/ than that as /ˈtokjo/ myself...

[Ser]

Anyway, I wouldn't be surprised if this is a thing; Cj clusters are pretty rare except before /u/.

Yeah, English phonotactics just don't work that way.

I remember my first week going to school in Canada. People gave me all sorts of weird looks when I was asking about some Mr. [ljɑŋ] ("Liang", it is pronounced [ljɑŋ] in Mandarin, and I thought it was okay to adapt it that way). How would I know that didn't obey English phonotactics, and that people here pronounce it [lɪ.ˈjɑŋ] instead?

[Imralu]

How would I know that didn't obey English phonotactics

=> How could/should I have known that I wasn't obeying English phonotactics ... ?

When I see foreign names, I automatically change my pronunciation now. For example, if it's a Spanish name I adopt a kind of miscellaneous foreign phonology (where /r/ is [ɾ]), and sometimes I'm not sure how your average uneducated monolingual Aussie would pronounce it anymore. For example, the Chinese name Leung. I would pronounce that [leʊŋ] but I have no idea how most Australians would think to pronounce that. Maybe L+EU+NG = /luːŋ/ or maybe LE+UNG /liːˈʌŋ/.

I also go into Italian restaurants and order, gnocchi if not [ɲɔkːi] then at least [njɔkʰi]. Usually the wait(e)r(ess) repeats it as [nɔki], which is fine, but once I had a confused waitress just look at me and then when I pointed she went "Oh, the [gəˈnɔtʃi]". And I like to drink chinotto, and MOST people who work in restaurants here, who aren't actually Italian, say [tʃəˈnɔɾəʉ]. I feel like asking "What would you recommend? The [spəˈdʒeɾi bɔləgˈniːz] or the [ləˈsægni]?

[Chagen]

I also go into Italian restaurants and order, gnocchi if not [ɲɔkːi] then at least [njɔkʰi]. Usually the wait(e)r(ess) repeats it as [nɔki], which is fine, but once I had a confused waitress just look at me and then when I pointed she went "Oh, the [gəˈnɔtʃi]". And I like to drink chinotto, and MOST people who work in restaurants here, who aren't actually Italian, say [tʃəˈnɔɾəʉ]. I feel like asking "What would you recommend? The [spəˈdʒeɾi bɔləgˈniːz] or the [ləˈsægni]?

I have never heard people butcher Italian loans THAT badly.

[Travis B.]

I also go into Italian restaurants and order, gnocchi if not [ɲɔkːi] then at least [njɔkʰi]. Usually the wait(e)r(ess) repeats it as [nɔki], which is fine, but once I had a confused waitress just look at me and then when I pointed she went "Oh, the [gəˈnɔtʃi]". And I like to drink chinotto, and MOST people who work in restaurants here, who aren't actually Italian, say [tʃəˈnɔɾəʉ]. I feel like asking "What would you recommend? The [spəˈdʒeɾi bɔləgˈniːz] or the [ləˈsægni]?

I have never heard people butcher Italian loans THAT badly.

Same here. The canonical loan pronunciations of spaghetti and lasagna that I am used to are /spəˈɡɜti/ > [spəːˈɡ̊ɜ̟ɾ̥i(ː)] and /ləˈzanjə/ > [ʟ̞əːˈzãːnjə(ː)]~[ɰəːˈzãːnjə(ː)], which, while showing their borrowed-ness, are far more reasonable adaptations of the original Italian words to English phonology.

[Ser]

How would I know that didn't obey English phonotactics

=> How could/should I have known that I wasn't obeying English phonotactics ... ?

"That" here is demonstrative pronoun, with some significant stress, it's not a conjunction. In other words, "How would I know [that] that didn't obey English phonotactics?", where "that" stands for "pronouncing <Liang> [ljɑŋ]".

When I see foreign names, I automatically change my pronunciation now. For example, if it's a Spanish name I adopt a kind of miscellaneous foreign phonology (where /r/ is [ɾ]), and sometimes I'm not sure how your average uneducated monolingual Aussie would pronounce it anymore. For example, the Chinese name Leung. I would pronounce that [leʊŋ] but I have no idea how most Australians would think to pronounce that. Maybe L+EU+NG = /luːŋ/ or maybe LE+UNG /liːˈʌŋ/.

It's [lœːŋ ~ lœwŋ] in Cantonese, if that helps. So you could try using your /lɵːŋ/ for it if you'd like to approximate the Canto...

[Chargone]

How would I know that didn't obey English phonotactics

=> How could/should I have known that I wasn't obeying English phonotactics ... ?

"That" here is demonstrative pronoun, with some significant stress, it's not a conjunction. In other words, "How would I know [that] that didn't obey English phonotactics?", where "that" stands for "pronouncing <Liang> [ljɑŋ]"....

not sure if i'm understanding you right there, but if i am, you'd want 'how was i to know that didn't obey ...' doesn't really matter what you're doing, so far as i can tell 'would' doesn't fit because it's talking about the past.

my explanation's bound to be missing something but i can't find a way to fit the word 'would' in that slot that works...

[finlay]

I don't think I've ever heard "Tokyo" pronounced with two syllables.

I have. But I live there.

Needless to say I usually hear the Japanese pronunciation of it, like [toː.kjoː], so it's not quite the same. I'm not sure whether it's two or three when I say it these days, actually; if it's three, the [ki] syllable is very short. I tend to pronounce it with an Englishy diphthong when I'm speaking English, anyway. I should conduct an informal survey of English speakers that I know here...

[clawgrip]

When I say Tokyo in English I retain /kj/ and pronounce it as two syllables, but the o sounds are blatantly English (Canadian), something like /ɔʊ/. When I speak in Japanese I use a Japanese /o/. Before I learned any Japanese I'm fairly certain that I pronounced it as three syllables, since I have never, ever heard an English speaker who knows no Japanese say it as anything other than the three-syllable /ˈto.ki.o/.

I think the main thing is that (and correct me if I'm wrong), /Cj/ doesn't tend to happen much in English aside from /Cju/, so other vowels tend to push the /j/ into becoming /i/.

[finlay]

Pretty much. I think English phonology tends to treat /ju/ as a diphthong rather than a CV sequence (the orthography certainly treats it that way). Tokyo as /tokjo/ in English isn't unpronounceable, it's just obviously not an English word.

Also, even though my accent is very light, I can get away with monophthongal /o/ in English because I grew up in Scotland, so I can sort of get away with pronouncing it the Japanese way in my English. What I find weird is when some non-rhotic English people use their THOUGHT/NORTH vowel for the Japanese long o, so they end up saying Torkyor or Orsaka. It's phonetically accurate, but sounds weird to me because the Japanese long o sounds like a GOAT vowel to me, and maps onto my phonological system that way.

[Nortaneous]

How many syllables does 'fjord' have?

[clawgrip]

Pretty much. I think English phonology tends to treat /ju/ as a diphthong rather than a CV sequence (the orthography certainly treats it that way). Tokyo as /tokjo/ in English isn't unpronounceable, it's just obviously not an English word.

This would make the most sense. If /Cj/ is not a valid sequence in English, we will avoid it as much as possible by replacing /j/ with /i/.

Also, even though my accent is very light, I can get away with monophthongal /o/ in English because I grew up in Scotland, so I can sort of get away with pronouncing it the Japanese way in my English. What I find weird is when some non-rhotic English people use their THOUGHT/NORTH vowel for the Japanese long o, so they end up saying Torkyor or Orsaka. It's phonetically accurate, but sounds weird to me because the Japanese long o sounds like a GOAT vowel to me, and maps onto my phonological system that way.

In my experience, the non-rhotic <or> ([ɔː], [ɔ]) is basically always what replaces Japanese [o] for (England) English speakers who speak Japanese but cannot shed their native accent. It sounds odd to me because [ɔː] doesn't exist anywhere in my phonological system (I am a victim of COT/CAUGHT, so in that range I have only [ɑ], the rhotacized [ɔ˞], and the aforementioned [ɔʊ], which I perceive as a more closed vowel than [ɔ]), or in the Japanese phonological system either. When I hear it in Japanese, I immediately perceive the speaker to be an English English speaker. I suppose an RP speaker identifies [ɔ] as being more similar to Japanese [o] than the closest RP alternative, [əʊ], and that's why they use it.

How many syllables does 'fjord' have?

I pronounce it with /fj/, and I think a lot of other people do too. I think when it is a stressed syllable, English-speakers are more tolerant about retaining the /Cj/ pronunciation. Kyoto seems more likely to be rendered as /kiˈo.to/, but /ˈkjo.to/ is not out of the question either.

[Gojera]

Yes, I was thinking that Arya vs. Ariya is an issue of unpredictable stress. It's easier to say as two syllables if you stress the second one, but then that sounds like "Are ya?"

[finlay]

Pretty much. I think English phonology tends to treat /ju/ as a diphthong rather than a CV sequence (the orthography certainly treats it that way). Tokyo as /tokjo/ in English isn't unpronounceable, it's just obviously not an English word.

This would make the most sense. If /Cj/ is not a valid sequence in English, we will avoid it as much as possible by replacing /j/ with /i/.

Also, even though my accent is very light, I can get away with monophthongal /o/ in English because I grew up in Scotland, so I can sort of get away with pronouncing it the Japanese way in my English. What I find weird is when some non-rhotic English people use their THOUGHT/NORTH vowel for the Japanese long o, so they end up saying Torkyor or Orsaka. It's phonetically accurate, but sounds weird to me because the Japanese long o sounds like a GOAT vowel to me, and maps onto my phonological system that way.

In my experience, the non-rhotic <or> ([ɔː], [ɔ]) is basically always what replaces Japanese [o] for (England) English speakers who speak Japanese but cannot shed their native accent. It sounds odd to me because [ɔː] doesn't exist anywhere in my phonological system (I am a victim of COT/CAUGHT, so in that range I have only [ɑ], the rhotacized [ɔ˞], and the aforementioned [ɔʊ], which I perceive as a more closed vowel than [ɔ]), or in the Japanese phonological system either. When I hear it in Japanese, I immediately perceive the speaker to be an English English speaker. I suppose an RP speaker identifies [ɔ] as being more similar to Japanese [o] than the closest RP alternative, [əʊ], and that's why they use it.

Probably. You see, I have something that's probably roughly halfway between RP and AmE, which basically involves [ɔ] for LOT/THOUGHT (ie cot/caught) and [o] (varying with a diphthong and usually longer) for GOAT - so the best way for me to map Japanese vowels onto my dialect is by interpreting /o/ as my LOT vowel and /oː/ as my GOAT vowel, which isn't strictly accurate but helps me keep them in check. One of the Japanese receptionists mentioned this a few weeks ago - English people normally have the most trouble with that particular vowel in Japanese, but also the length thing in general, because our vowels are divided up by quality, and length is a secondary feature. So it's possible to readjust your mind, since at the end of the day they have fewer vowels than we do, but learning to hear it is still difficult.

/u~uː/ is another Japanese vowel that I've heard a lot of people having trouble with - especially when pronouncing Shinjuku, which tends to end up as /ʃɪnˈdʒuku/ in the English of many people, and the stressed /u/ is usually a diphthong like [ʉw]. The best way I have of countering that is to stress it on the first syllable, like /ˈʃɪndʒuku/, and then since my /u/ is very rarely a diphthong it's not a problem anymore.

[clawgrip]

our vowels are divided up by quality, and length is a secondary feature. So it's possible to readjust your mind, since at the end of the day they have fewer vowels than we do, but learning to hear it is still difficult.

In most (all?) varieties of English, vowel length is not phonemic, so learning to differentiate length gives a lot of people trouble.

/u~uː/ is another Japanese vowel that I've heard a lot of people having trouble with - especially when pronouncing Shinjuku, which tends to end up as /ʃɪnˈdʒuku/ in the English of many people, and the stressed /u/ is usually a diphthong like [ʉw]. The best way I have of countering that is to stress it on the first syllable, like /ˈʃɪndʒuku/, and then since my /u/ is very rarely a diphthong it's not a problem anymore.

The [u͍] vowel in Japanese is a relatively rare vowel sound among the world's languages, so speakers of most languages have trouble getting it right. But since there are so few vowels in Japanese, if you say or even [ʉw], your meaning will still be clear.

Merging this topic back into the OP's topic, one of the the most troublesome consonant clusters in Japanese for English-speakers is [ɺ̺j]. This is one of the last sounds people learn to get right (aside from maybe /n̩V/, e.g. <kin'en> [kʲĩː(j)ɛɴ]) and for English speakers who can't say it properly, it invariably results in /rij/, e.g. <ryokan> /ɺ̺jo.kaɴ/ becomes /ri.jo.kan/.

[Soap]

Yeah I have trouble with that too. At least it can only precede /o/ and /u/, unlike Russian. I would just say /riwkju/ for Ryukyu (not that I'll likely ever have to say it in Japanese in real life).

[clawgrip]

Yeah I have trouble with that too. At least it can only precede /o/ and /u/, unlike Russian. I would just say /riwkju/ for Ryukyu (not that I'll likely ever have to say it in Japanese in real life).

There's also rya.

[finlay]

our vowels are divided up by quality, and length is a secondary feature. So it's possible to readjust your mind, since at the end of the day they have fewer vowels than we do, but learning to hear it is still difficult.

In most (all?) varieties of English, vowel length is not phonemic, so learning to differentiate length gives a lot of people trouble.

/u~uː/ is another Japanese vowel that I've heard a lot of people having trouble with - especially when pronouncing Shinjuku, which tends to end up as /ʃɪnˈdʒuku/ in the English of many people, and the stressed /u/ is usually a diphthong like [ʉw]. The best way I have of countering that is to stress it on the first syllable, like /ˈʃɪndʒuku/, and then since my /u/ is very rarely a diphthong it's not a problem anymore.

The [u͍] vowel in Japanese is a relatively rare vowel sound among the world's languages, so speakers of most languages have trouble getting it right. But since there are so few vowels in Japanese, if you say or even [ʉw], your meaning will still be clear.

Merging this topic back into the OP's topic, one of the the most troublesome consonant clusters in Japanese for English-speakers is [ɺ̺j]. This is one of the last sounds people learn to get right (aside from maybe /n̩V/, e.g. <kin'en> [kʲĩː(j)ɛɴ]) and for English speakers who can't say it properly, it invariably results in /rij/, e.g. <ryokan> /ɺ̺jo.kaɴ/ becomes /ri.jo.kan/.

Japanese /u/ is very similar to the one in most British accents, in that it's fronted compared to cardinal . I don't know of any major English accents with anything like cardinal in them. And the secret to Japanese /rj/ (IMO) is to think of it as [lj], which exists in some English accents, or at least isn't so outlandish (their /r/ is more similar to our /l/ than our /r/, anyway). You can sort out the details, like what phonetic value the Japanese /r/ actually has, later. You've labelled it as [ɺ] (although you need to sort out what brackets you're using), which I guess is the normal way, but it's not rigidly that every time they say anything, if you see what I mean. I think it has other possibilities.

(to me ry sounds like y, anyway...)

As for discussions of English, it's such a widely spoken language that you can hardly ever make such generalizations about all English dialects. Australian English is the most obvious one with pure length contrasts, such as "cunt” [kant] vs "can't" [kaːnt] among others. The traditional IPA used for RP uses length marks, and length is important if you want to pronounce RP English 'properly', but it's not the primary distinguishing feature. Length for me is almost completely allophonic, although there is something morphological (allophone conditioned by the morpheme boundary) going on with brood [brʉd] and brewed [brʉːd], and when I pronounce cot and caught together, the second one is the same vowel quality but slightly longer (this is because I know there's supposed to be a difference). These are basically the scottish features in my accent that nobody actually notices because it's mostly only linguists that know about them. But I'm used to listening to English accents with long and short vowels, even if, as I say, they're not the primary distinguishing factor between the two.

Plus, fuck it, I have a well trained ear for this shit. I just need to actually learn some useful Japanese.

[spats]

Japanese /u/ is very similar to the one in most British accents, in that it's fronted compared to cardinal . I don't know of any major English accents with anything like cardinal in them.

I don't find the Japanese /u/ to be that hard to "do", but neither would I say that it's similar to the BrE or AusE /u:/ vowel in any other sense than it's a high vowel falling somewhere between cardinal /u/ and /i/. My impression was that BrE and AusE /u/ were strongly endolabial, fairly central, and had a strong glide component. Japanese /u/ is near-back, somewhat exolabial, and has no glide.

[clawgrip]

Japanese /u/ is very similar to the one in most British accents, in that it's fronted compared to cardinal . I don't know of any major English accents with anything like cardinal in them.

I think you can find this in various Caribbean English accents.

And the secret to Japanese /rj/ (IMO) is to think of it as [lj], which exists in some English accents, or at least isn't so outlandish (their /r/ is more similar to our /l/ than our /r/, anyway). You can sort out the details, like what phonetic value the Japanese /r/ actually has, later. You've labelled it as [ɺ] (although you need to sort out what brackets you're using), which I guess is the normal way, but it's not rigidly that every time they say anything, if you see what I mean. I think it has other possibilities.

I realize that although I used [ɺ], Japanese /r/ can be also be realized as [ɽ], [ɾ], or even [r]. I probably should have just used /r/ and been done with it.

(to me ry sounds like y, anyway...)

There's a definite /r/ in there. Try practicing with some minimal pairs: 水浴 <suiyoku> 'bathing' 水力 <suiryoku> 'hydraulic power', 略す <ryakusu> 'to abbreviate', 訳す <yakusu> 'to translate', 領事 <ryōji> 'consul', 用事 <yōji> 'task'.

As for discussions of English, it's such a widely spoken language that you can hardly ever make such generalizations about all English dialects. Australian English is the most obvious one with pure length contrasts, such as "cunt” [kant] vs "can't" [kaːnt] among others. The traditional IPA used for RP uses length marks, and length is important if you want to pronounce RP English 'properly', but it's not the primary distinguishing feature. Length for me is almost completely allophonic, although there is something morphological (allophone conditioned by the morpheme boundary) going on with brood [brʉd] and brewed [brʉːd], and when I pronounce cot and caught together, the second one is the same vowel quality but slightly longer (this is because I know there's supposed to be a difference). These are basically the scottish features in my accent that nobody actually notices because it's mostly only linguists that know about them. But I'm used to listening to English accents with long and short vowels, even if, as I say, they're not the primary distinguishing factor between the two.

I knew I was probably wrong by suggesting that all dialects lack phonemic vowel length, However, I think it’s probably safe to say that in the cases it does exist, it’s only contrastive in a small percentage of the total vowel inventory, and is not a fundamental feature of the language. Because of this, English speakers on the whole are not at first accustomed to consciously recognizing or realizing phonemic vowel length on any and all vowels, even if it occurs in their dialect to some extent. Length is not phonemic in my accent though, so I can't say with absolute certainty. I mainly base it on my experience listening to people not correctly producing long vowels. It seems that English speakers initially will impose English-style stress points on Japanese words and lengthen whatever vowel that stress point falls on.

I don't find the Japanese /u/ to be that hard to "do", but neither would I say that it's similar to the BrE or AusE /u:/ vowel in any other sense than it's a high vowel falling somewhere between cardinal /u/ and /i/. My impression was that BrE and AusE /u/ were strongly endolabial, fairly central, and had a strong glide component. Japanese /u/ is near-back, somewhat exolabial, and has no glide.

I am not sure I agree with your description. In my (albeit somewhat limited) knowledge of AusE and BrE pronunciation, it seems to me that Australian /u/ is very central, while British /u/ is further back, and Japanese /u/ falls somewhere in the middle, though a little further back and closer to BrE, as finlay was implying. Nevertheless, as you say, there is no glide, and English speakers often have trouble removing glides from their vowels, and I know that a British friend of mine had quite a time trying to pronounce Japanese /u/ properly, despite its similarity to the vowel in his own speech.

I have to say though that I’m not sure where you get the idea that Japanese /u/ is exolabial, and that Australian and British /u/ are endolabial. It seems like you’ve got it backwards. I can’t speak with complete confidence about Australian and British /u/, but I do not agree with your assessment of Japanese /u/ being exolabial, and I have never seen anyone else who agrees either.