For many dialects of English (including my own) multiple historical lexical sets are merged into one "FLEECE" set (this diaphoneme can be represented with IPA /iː/).
I've read about the basics of the "MEET-MEAT" merger, but apparently the situation is more complicated than I understood and involves a greater number of lexical sets than just two. Furthermore, the orthography does not always indicate which lexical set a word belongs/belonged to.
According to Wikipedia, the "meet-meat" merger actually has acted on the the following, originally distinct, lexical sets:
In accents with the distinction, the vowel /ɪə/ is usually represented by the spellings ea and eCe, as in neat and complete, and the vowel /ɛɪ/ is usually represented by the spellings ei and ey, as in receive and key, and the vowel /iː/ is usually represented by the spellings ee, ie and iCe as in feet, thief and suite, as well as plain e in the monosyllabic words be, he, me, she, the (when stressed), we and ye.
So according to Wikipedia, some dialects make, or at least made, distinctions that overall require at least four lexical sets:
FEET /iː/: mostly from Middle English monophthongal /eː/ (which in turn mostly corresponds to the Old English long monophthong /ē/); sometimes corresponds to orthographic <ie> for reasons unclear to me
- <ee> (feet)
- <ie> (thief)
- <i>, apparently in recent-ish borrowings from foreign languages (suite)
TEAM~CREAM /ɪə/: from vowels in English and French that originally had a lower vowel quality, but were still long
- Old English long diphthong /ēa/ (team)
- Old English long monophthong /ǣ/ (apparently, if spelling is an accurate guide)
- French /ɛ/, which seems to have been lengthened as a rule when taken into English (cream)
EAT /ɛɪ/: from original short /e/ lengthened by Middle English open syllable lengthening
- Old English short vowel /e/ subsequently lengthened in an open syllable (meat, eat)
Apparently distinguishing CREAM and EAT is rare; most dialects (and the orthography) have merged the above two categories into a single MEAT lexical set.
KEY~RECEIVE /ɛɪ/: from /ɛɪ/ diphthongs, apparently
- from Old English /ǣġ/ (key) (possibly irregular development?)
- from /ɛɪ/ diphthong in French/Norman (receive and similar words)
Wikipedia is very unclear about which dialects have this as a distinct lexical set, and how dialects that don't merge it with other lexical sets. A World Heritage Encyclopedia entry hosted by Project Gutenberg suggests that for (at least some) Yorkshire accents, KEY is actually not a distinct lexical set from EAT.
[eɪ] may take the place of /iː/, especially in words such as key, meat, speak.
I feel like the Wikipedia article didn't completely explain the historical development or the synchronic distribution of forms in different dialects, so I was wondering if anyone is more familiar with this topic (or knows how I can access more detailed data about the pronunciation of different words in non-merged dialects.)
My main question is: in dialects that generally don't exhibit a complete FLEECE merger, are there any neutralizations between the various lexical sets in particular environments?
One particular example I've been wondering about: does anybody know which lexical set the stressed vowels in words like idea, urea and diarrh(o)ea belongs (or belonged) to for non-merged speakers? That is, a phonetically lengthened, but orthographically single <e> located before another vowel in a Latinate word. (I have not heard of any English dialects that have a general distinction between <e>, <ae> and <oe> in words derived from Latin, so I'd assume the same pronunciation patterns apply to these digraphs as for the single letter.)
Since the vowel here seems to be lengthened, not originally long, I would guess it to be in the EAT set. But I also thought that perhaps learned words like this might tend to be influenced by the pronunciation used in "standard" dialects, which tend to have a complete merger.
I also wondered because in Received Pronunciation, FLEECE when followed by a schwa behaves in a special way: it may be "smoothed" into a centering diphthong /ɪə̯/ (lexical set NEAR). I get the impression that smoothing of a vowel-schwa sequence to a diphthong is even more common for the unstressed happY vowel followed by schwa, as in the word arboreal.
Another situation where vowel reduction (and neutralization of contrasts) seems likely to me is in words like create, where the vowel is unstressed and comes directly before a stressed vowel (in Received Pronunciation, "create" has the "happY" vowel /ɪ/).
In fact, I have more questions about non-merged dialects, but I think that should be enough for now; I'll wait first to see if anyone can help me with this one.