knees niece merger

Lilly   Mon Jul 09, 2007 5:46 pm GMT
I've heard that people in Chicago have KNEES NIECE merger. Is it true?
Both are pronounced: [ni:s]
Do you have this merger?
Travis   Mon Jul 09, 2007 5:55 pm GMT
This is not an actual merger at least here in Milwaukee (not too far north of Chicago), because a vowel length distinction exists:

"niece" [nis]
"knees" [ni:s]

Also, "knees" will commonly have [z] here if the next word starts with a vowel.

The matter is that vowel length allophony is still active here, and "niece" underlyingly has /s/ while "knees" still underlyingly has /z/. Consequently, the two are merged by final devoicing if the next word does not start with a vowel (or in cases, an approximant or lenis obstruent), but the vowel lengths resulting from vowel length allophony are not affected by the devoicing.
Guest   Mon Jul 09, 2007 6:11 pm GMT
What about "eyes" and "ice"? A Meixcan teacher I had would always say "ice" instead of "eyes".
Travis   Mon Jul 09, 2007 6:26 pm GMT
They're not homophones for me either, as they are:

"eyes" [a:Is]
"ice" [@Is]

Note that the difference in the diphthong's quality here is due to Canadian Raising, which is conditioned by the underlying distinction between /s/ and /z/.

And likewise, "eyes" often has [z] for me before a word starting with a vowel or, less commonly, an approximant or lenis obstruent.
furrykef   Mon Jul 09, 2007 7:02 pm GMT
Hmm, I had never heard of this before. Pronouncing "knees" as "niece" (with an "s" sound instead of a "z" sound) would confuse me, no matter if there is a vowel length distinction between them.

- Kef
Travis   Mon Jul 09, 2007 7:13 pm GMT
>>Hmm, I had never heard of this before. Pronouncing "knees" as "niece" (with an "s" sound instead of a "z" sound) would confuse me, no matter if there is a vowel length distinction between them.<<

This is largely an Upper Midwestern thing, a substratum feature left behind by the immigrant languages once spoken here (e.g. German and Polish). It's really not surprising that one would not encounter it if one lived outside an area with strong substratum influence from languages either with final devoicing or lacking coda voiced obstruents.
Travis   Mon Jul 09, 2007 7:27 pm GMT
I have to say that for me it is not confusing, but that's because I naturally listen for the lengths of vowels, not just their voicing, when determining lenis-fortis values of obstruent phonemes. Often vowel length says more about such than the actual voicing of obstruents, as while voicing is subject to assimilation or devoicing, preceding vowel length is not.

At the same time, though, this goes the other way, as I often find it very hard to understand people, generally non-native English-speakers, who get vowel allophony wrong. They can provide the expected voicing of obstruents, and yet I will perceive them differently due to the incorrect vowel lengths.

Another factor is glottalization, as postvocalic fortis plosives in many English dialects are glottalized *but*, at least here, devoiced lenis plosives lack such glottalization in postvocalic positions. This ties in with vowel length, and often makes it easier to determine the actual fortis-lenis values of postvocalic plosives than with vowel length alone. Again, lack of such glottalization can be confusing, as I may be liable to perceive voiceless postvocalic plosives as lenis without it.