Are there minimal pairs between vowels of normal length such as
a and vowels of long length such as
Australian English has true phonemic length distinctions.
- ferry /feɹi/ vs fairy /feːɹi/
- Manning (name) /mænɪŋ/ vs manning /mæːnɪŋ/
A classic example is that virtually all dialects distinguish [bid] "bid" and [bi:d] "bead". Also, dialects often distinguish [bit] "bit" and [bi:t] "bid". A "minimal pair" is a pair of words whose phonetic valued differ only in the presence of A vs. B in the phonetic output. To rephrase, "minimal pair" is a property of phonetic outputs, not underlying phonological analysis. It is a well-known problem with the concept of "minimal pair" that the minimality of the pair is crucially determined by the granularity of the transcription, so while [bid] vs. [bi:d] is "okay" as a phonetic transcription, [bɪd] / [bi:d] is even better (and also not minimal w.r.t. length), and [bɪˑt] / [bɪˑjt] is best (though perhaps dialect dependent). Then we have the contrast [bɪˑt] "bid", [bɪt] "bit", [bɪˑjt] "bead" and [bɪjt] "beat/beet". The underlying contrast in the vowels is length, which is realized phonetically via diphthongization. The underlying consonant contrast is voicing, realized phonetically as lengthening of the vowel.
My "informants" of Australian English had a vowel quality contrast to go with their length contrast. The Wiki article says "some speakers" and the data are notated with a "citation needed" flag indicating that there is a diphthongization difference to go with the assumed length difference. This can only be resolved with some concrete acoustic evidence. Fletcher et al. look into the "hud / hard" difference, unfortunately their measurements only give a single generic formant value of "the vowel". They do show that there is a large correlation between vowel choice and duration, and a small and speaker-dependent difference in formant value.
In my west coast dialect, there are minimal pairs like "whore" [hɔɹ] and "horror" [hɔ:ɹ], "tear" [tɛɹ] and [tɛ:ɹ]. To validate the claims about transcription, we'd have to have an independent way to establish how much of the speech signal is [ɹ] and how much is vowel. If the rhotic is longer in the orthographically bisyllabic forms, thus we really have [hɔɹ] vs. [hɔ:ɹˑ], then this would not be a minimal pair, it would be like the case of "beat" and "bit", with length and quality co-dependent.
Yes, though it depends on the variety. For example in Scottish English, vowel length does not have minimal pairs since it is determined by "Aitken's law". In Standard Southern British English (SSBE) however there are minimal pairs for example:
- For /ɪ/ vs. /iː/: bit vs. beat; lick vs. leek; hit vs. heat etc.
- For /æ/ vs. /ɑː/: cat vs. cart (though this wouldn't hold up in rhotic accents)
Also note the qualitative difference in the short and long vowels, it's not just a vowel length difference, so some might argue that it's not actually vowel length which forms the minimal pair but the quality of the vowel.
Hope this helps.
Non-rhotic varieties of English have a number of phonemic contrasts in which vowel duration is pretty important. It might be a primary cue to the vowel contrast in some varieties, as exemplified above. Perhaps English will develop more durational minimal pairs.
For example, the vowel in "hut" is often a short version of the vowel in "heart" or STRUT vs. HEART in lexical sets, though this is not conveyed in traditional transcriptions.
Pairs distinguished by duration (alone, or primarily) are increasingly met for
"piss" vs. "pierce" (KIT vs. NEAR in closed syllables)
"led" vs. "laird" (DRESS vs. SQUARE in closed syllables)
Scottish English has short vs. long in near-minimal pairs like "need" vs. "kneed", "brood" vs. "brewed", distinguished purely by vowel duration. This is pretty much only possible for the vowels /i/ and /u/ before a /d/, where the /d/ is either tautomorphemic (for the short one) or a suffix (for the long one). There are also some words where a long vowel occurs unexpectedly, like "dude" and "lewd".
Given that the distribution of the vowel duration is mostly predictable from the status of the /d/, strictly speaking the short-long duration difference is not phonemic, and is one aspect of the so-called Scottish Vowel Length "Rule".