----- Original Message ----

> From: Roger Mills <[log in to unmask]>
> You're comparing apples (Engl. phonology) and oranges (Jap. phonology). [ts] is 
> a cluster in Engl. because in native words it (1) only occurs over a syllable or 
> morpheme boundary, (2) never initially. I doubt we could find a minimal pair 
> /t/:*/ts/ except maybe itty-bitty vs. itsy-bitsy** Ergo, not a _phoneme_ but a 
> cluster.  In Japanese, [ts] is indeed an allophone of the phoneme /t/ preceding 
> /u/, and pronouncing it so is probably as natural and automatic for a Japanese 
> speaker as aspirating /p t k/ is for us.  

I found this confusing. For one thing, many linguists say that a consonant cluster *can't* occur over a syllable boundary (that is, by that definition, /ts/ in /c&t$sUp/ is not a consonant cluster because /t/ and /s/ are in different syllables).

It's very easy to find a /t/~/ts/ minimal pair: Take just about any noun or verb that ends in /t/ and add the plural/3p-sing affix <s>: /hIt/~/hIts/, /c&t/~/c&ts/, etc. To my understanding, a minimal pair doesn't demonstrate that the two things in contrast are different phonemes, it demonstrates that the two things in contrast represent one or more different phonemes. <hit> vs <hits> COULD demonstrate that [t] and [ts] represent /t/ and /t_s/ if there were other evidence of /t_s/, but the standard and more defensible analysis in English is that <hits> contains all the phonemes of <hit> *plus* /s/ (that is, that /s/ contrasts with null rather than that /t_s/ contrasts with /t/*). The relevant point, though, is that these examples all contain multiple morphemes. The "never initially" is not particularly relevant for English because we don't have any prefixes that are a single consonant.

> Similarly Engl. [tS, dZ] _are_ phonemic//not clusters because they can occur 
> initially, medially, finally and there are lots of minimal pairs. [tS] as we've 
> been discussing, can also occur over a word boundary.

[N] cannot occur initially in English; in standard dialects, [h] can only occur initially. There are no minimal pairs contrasting these. And yet it's only the most obstreperous of linguists who would posit /h/ with allophones [h] and [N]. On the other hand, [st] can appear initially (stop), medially (master), or finally (best); and there are plenty of minimal pairs (stop vs top, master vs matter, best vs bet); but [st] is not a manifestation of a single phoneme /s_t/ in English (under standard analyses). Again, constrastive pairs indicate that different phonemes are involved, but they don't definitively illustrate how many phonemes are on each side of the contrast.

To go back to the original issue, though, a question for the list: If /t_s/ is not in fact an English phoneme, then how do we treat "tse-tse," "tsunami," etc. (including the contrastive <tsars>/<SARS>), when said with [ts]? Is Gary correct that, in English, /ts/ is an initial consonant cluster in these loan words? I'm convincing myself that, for my pronunciation of these words at least, he is.

-- Paul

* I realize now I did indeed get the "catch it" example backwards, the point of it being that <catch it> contains /t_S/ while <cat shit> contains /tS/. /t_S/ is a single phoneme in the former, and so it will not separate: People will say either [k&t_S$It] or [k&$t_SIt], but not [k&t_h$SIt]. If they move any part of the /t_S/ across the syllable boundary, they'll move the entire thing, indicating that they mentally represent it as a single phoneme.