s and z before sh

Johnny   Wed Oct 08, 2008 8:16 pm GMT
Do native speakers connect words like this?

This shirt -> Thish shirt
Whose shirt is that? -> I can't figure this out, I think it's "whouge shirt", where whouge rhymes with rouge.

Thanks.
Lazar   Wed Oct 08, 2008 8:26 pm GMT
Yes. (Are you familiar with IPA and X-SAMPA? If not see, http://en.wikipedia.org/wiki/X-SAMPA .) In my speech, alveolar fricatives consistently assimilate to following postalveolar fricatives:

/s.S/ < [S.S] e.g. "this shirt" [%DIS "S3`t]
/z.S/ < [Z.S] e.g. "Joe's shirt" ["dZ7UZ "S3`t]
/s.Z/ < [S.Z] e.g. "this genre" [%DIS "ZA:nr\@]
/z.Z/ < [Z.Z] e.g. "whose genre" ["hu:Z "ZA:nr\@]

Although the last two cases are marginal, so they don't come to me quite as naturally. But yeah, I think all native speakers would do this - it would strike me as practically impossible to say [%DIs "S3`t] or ["dZ7Uz "S3`t].
Travis   Wed Oct 08, 2008 10:09 pm GMT
My dialect is different in this regard in that such assimilation only necessarily occurs if the following consonant is palatalized, with palatalization being conditioned by either a following /u/, /ʊ/, /ər/, and sometimes /o/ or /ɛ̞/ (which I previously transcribed as /ɛ̯æ/), or by being a sibilant followed by a plosive other than /k/ or /p/, or by /m/, /n/, /r/, /l/, or /w/. If it is not, normally such assimilation will *not* occur except maybe occasionally in very uncareful speech. Note though that this is separate from voicing assimilation, which practically occurs without exception across word boundaries not separated by a pause in the dialect here (but which may be complicated by things such as final devoicing and /dʒ/ always being voiceless). Anyways, as a result of this, I normally have:

"this shirt" [ˈ(d/d̪/ð)ɪʃʲˈʃʲʁ̩ʔ(t)]
"Joe's shirt" [ˈd̥ʒ̥oː(ʃʲ/ʒ̥ʲ)ˈʃʲʁ̩ʔ(t)]

but:

"this genre" [ˈ(d/d̪/ð)ɪsˈ(d̥ʒ̥/ʒ̥)ãː(n)ʁəː]
"whose genre" [ˈ(h)uː(s/z̥/z)ˈ(d̥ʒ̥/ʒ̥/ʒ)ãː(n)ʁəː]
Johnny   Wed Oct 08, 2008 10:50 pm GMT
Thanks. Cool, so I guessed right, Lazar. :)
<<it would strike me as practically impossible to say [%DIs "S3`t] or ["dZ7Uz "S3`t]. >>
Yeah, I couldn't say that either, so I tried to think of a natural way to say it, and I guessed, lol. I can't speak at a decent speed yet, because of certain word connections. In accent reduction courses they only explain some instances of assimilation, like t + y = ch, but not the one I asked about.

Another example I asked about recently was the difference between, say, "I like them" and "I liked them". Or "organize things" and "organized things"... I really can't notice any difference. I guess this might be called reduction in consonant clusters, and again, I don't think there is much emphasis on such features in the ESL field, unfortunately.

Travis, your dialect is amazing, as expected, lol. The way you always describe it makes it seem very different from other dialects, but that might be because you seem to be the only one who describes their dialect in such a precise way, so it seems it has more features by comparison... but the others might be as complex as yours, in the end, if we tried to analyze them the same way.
By the way, do you even say things like "om monday" instead of "on monday"? That kind of assimilation doesn't seem to be common in the US (even though some dialects definitely have it, perhaps most Ebonics varieties, for example), while it seems much more common in the UK.
Johnny   Wed Oct 08, 2008 10:59 pm GMT
<<that might be because you seem to be the only one who describes their dialect in such a precise way>>

Interesting thing about singular "they". I guess that should have been "his dialect" anyway, even if it is supposed to be generic, since it comes after "the one who describes". If you were a female, it would have been "the only one who describes her dialect". How weird, lol. Just some thoughts as a non-native speaker who always uses singular they but felt it wasn't alright in this case. :)
Travis   Wed Oct 08, 2008 11:15 pm GMT
>>Travis, your dialect is amazing, as expected, lol. The way you always describe it makes it seem very different from other dialects, but that might be because you seem to be the only one who describes their dialect in such a precise way, so it seems it has more features by comparison... but the others might be as complex as yours, in the end, if we tried to analyze them the same way.<<

In my examples, the only things which were at all really markedly atypical for an NAE dialect were the uvular rhotics, actually. Other than that, though, my examples were really not all too atypical for examples from NAE dialects.

For instance, such palatalization is seemingly actually a normal feature of most NAE dialects (even though my dialect does have palatalization in places that others lack it). Most people never transcribe it, but from actually listening to NAE speakers not from here, it most certainly is practically always there if I listen for it.

Likewise, as for voicing, the pattern of voicing found in the dialect here is really only atypical for an NAE dialect in that it has optional final fortition of /z/, /ʒ/, /dʒ/, /f/, and less frequently /d/, while most NAE dialects only have phonetic final devoicing or half-voicing, and /dʒ/ is always voiceless in it. Other than that, it is quite normal for an NAE dialect to have voicing assimilation across word boundaries not separated by a pause and to devoice or half-voice final consonants not between vowels or adjacent to a lenis consonant across the word boundary in question.

In this kind of way, while my own dialect does most definitely have its strange points, quirks, and potentially-intelligibility-breaking-to-the-naive (read: non-North American) sound changes, it really is not as strange for an NAE dialect as one might think. In reality it is certainly far closer to most non-East Coast Northern NAE dialects than to any other English dialects. And yes, much of it is really just how I transcribe things, as it is far narrower than most people do, to the point that it certainly looks strange to those used to more typical transcription conventions, which typically omit things such as allophonic vowel length, vowel nasalization, preglottalization, palatalization, assimilation across word boundaries, final devoicing, voiceless lenis consonants, and so on.

>>By the way, do you even say things like "om monday" instead of "on monday"? That kind of assimilation doesn't seem to be common in the US (even though some dialects definitely have it, perhaps most Ebonics varieties, for example), while it seems much more common in the UK.<<

Oh, that is normal non-careful pronunciation for me, actually. I only do not do so if I intentionally force it, as in rather careful speech, as it is such a basic phonological rule to me. I am not so sure though whether such is actually that uncommon in NAE, even though in many dialects it does seem to be not as consistent as in my own.
Travis   Wed Oct 08, 2008 11:16 pm GMT
>>/z/, /ʒ/, /dʒ/, /f/, and less frequently /d/<<

should be "/z/, /ʒ/, /dʒ/, /v/, and less frequently /d/" above.
Johnny   Thu Oct 09, 2008 12:35 pm GMT
Yes, it seems your transcriptions are way too precise! LOL. I didn't even know what palatalization was, and after looking it up I realized it's a normal feature... but it's too complicated for me to analyze, as a non-native speaker and non-expert. From Wikipedia:
<<As an example, compare the /k/ of English key with the /k/ of coo, or the /t/ of tea with the /t/ of took. The first word of each pair is palatalized, but few English speakers would perceive them as distinct.>>
I notice a difference in production because palatalization is what comes natural, but as they say, I can't notice any difference in perception.

Anyway, could you tell me more about this part? I didn't understand much:
<<...while most NAE dialects only have phonetic final devoicing or half-voicing, and /dʒ/ is always voiceless in it. Other than that, it is quite normal for an NAE dialect to have voicing assimilation across word boundaries not separated by a pause and to devoice or half-voice final consonants not between vowels or adjacent to a lenis consonant across the word boundary in question.>>

What I sometimes notice is the final devoicing of /v/ and /z/, so you have something like "fife" instead of "five" and "ice" instead of "eyes", but the preceding vowel length is still long, as if followed by a voiced consonant, so pairs like "eyes" and "ice" for those who devoice final z's completely are only distinguishable by vowel length.
As for voicing assimilation across word boundaries, I don't understand exactly what you mean, but I was thinking of something like "messed up", where st+up comes out with a half voiced t, almost like s+dup... But that's just what I hear, remember I am not a native speaker.
Travis   Thu Oct 09, 2008 2:34 pm GMT
>>What I sometimes notice is the final devoicing of /v/ and /z/, so you have something like "fife" instead of "five" and "ice" instead of "eyes", but the preceding vowel length is still long, as if followed by a voiced consonant, so pairs like "eyes" and "ice" for those who devoice final z's completely are only distinguishable by vowel length.<<

This is definitely true when one has actual final fortition, as does show up in the dialect here. For me at least, it is not uncommon for there to be causes where two words are realized such that the only thing distinguishing them is vowel length and or preglottalization.

However, though, in most English dialects there is no final fortition but rather just phonetic final devoicing (which is often prevented by whatever the next word starts with), such cases generally do not actually show up. The thing is that while people commonly think of pairs such as [t] and [d] or [s] and [z] as just differing in voicing, in reality they do not differ in voicing alone. Rather, [d] and [z] are realized differently from their counterparts [t] and [s] aside from actual voicing per se. Hence the pure devoicing of [d] and [z] does not result in [t] and [s] but rather [d̥] and [z̥], which are phonetically shorter than [t] and [s] to say the very least. And for the record, sounds like [d̥] and [z̥] are referred to as being voiceless lenis, in contrast to voiceless fortis sounds like [t] and [s].

>>As for voicing assimilation across word boundaries, I don't understand exactly what you mean, but I was thinking of something like "messed up", where st+up comes out with a half voiced t, almost like s+dup... But that's just what I hear, remember I am not a native speaker.<<

What I was referring to is that in connected speech it is normal for two adjacent obstruents across word or morpheme boundaries to undergo voicing assimilation in NAE dialects, so that one causes the other to be devoiced or voiced. In most cases it seems that devoicing is favored over voicing except in particular cases where /d/ comes into contact with /t/ and both are reduced to a tap together. Take in mind what I was referring to above about voiceless lenis consonants - just because /d/ is devoiced in such a fashion does not mean that it is realized as [t] but rather that it is realized as [d̥] - and likewise such has no affect upon aspiration, preglottalization, or like. Such is why such is less noticable than one might think, as no merger of fortis/lenis phoneme pairs actually occurs as a result of such.