What are near-minimal pairs? How are they different from minimal pairs?
2Please don't edit your question into something new after it's already been answered.– Draconis ♦Feb 16 at 1:27
Sure. English "hit" and "hot" are near-minimal pairs, differing only in one phoneme, but the
/h/ is realized differently because of that (more like
[ç] in the first one).
It's not always possible to find a true minimal pair to distinguish phonemes, so near-minimal pairs can sometimes be necessary. You just need to make sure that whatever differs between them isn't likely to be causing allophony.
1You're using the term "phonemes can be in minimal pairs" wrong. Minimal pairs are one way we use to distinguish phonemes -- the really important sounds that speakers pay attention to -- from allophones -- insignificant variations on phonemes that speakers ignore. If you write out the words in IPA (if you can't do that, learn), you'll see that /h/ has a different allophone before every vowel it precedes.– jlawlerFeb 15 at 17:48
You are not using the term "phonemes can be in minimal pairs" at all, as far as I can see. Ninja edit?– vectoryFeb 15 at 19:52
1hit and hot are near minimal pairs?? The only difference is i and o. These are classic example of a minimal pair. hit=hɪt and hot=ˈhɑt [AmE] and hot=ˈhɒt [BrE]. Any of the vowels in these words make any two of them minimal pairs due only to the middle letter: hat, het, hit, hot, hut.– LambieFeb 15 at 19:54
1The English phoneme /h/ has 13 or so allophones -- one for every different vowel it precedes.– jlawlerFeb 15 at 20:16
2@Lambie I should clarify: a near-minimal pair for the difference between [ç] and [h]. They are of course a minimal pair for the difference between /ɪ/ and /ɔ/.– Draconis ♦Feb 15 at 20:38
A minimal pair is two distinct utterances which differ in the presence of exactly one surface phone, for example [væt] and [fæt]. The premise is that you first take note of various words as they are actually pronounced, so you have a record of the phones in the language. If you have a minimal pair, you have compelling evidence that the two phones derive from distinct phonemes, and are not allophones of a single phoneme.
A near-minimal pair is a pair of words which differ in more than one set of phones, where the number of differing phones is at least two, and hopefully not more than two. Near-minimal pairs are sometimes called on in cases where there are few or no minimal pairs to demonstrate the phonemic status of a phonic difference.
There is often chicken-egg confusion over the status of minimal pair in phonology. One does not know what the phonemes of a language are until one has performed a distributional analysis of phonetic segments, and that analysis can tell you that phonetic sounds X and Y are in complementary distribution (meaning that they can be treated as allophones of a single phoneme), or that X and Y are not in complementary distribution (are not allophones of a single phoneme). This then (supposedly) justifies positing certain allophonic rules, if the sounds have the requisite distribution. Sometimes, the evidence for complementary distribution is almost good enough, which is where near-minimal pairs become important (they can be used to establish phonemic contrast).
The crucial thing to remember is that one looks for minimal pairs in the phonetic record: it is not a hybrid of looking in the phonetic record but also looking at the abstract underlying form.
Everything that occurs in the phonetic record is an allophone (or, a phone). Allophone refers to the relation between a surface phone and whatever phoneme is is abstractly. [s] is an allophone of /s/, [p] and [pʰ] are allophones of /p/. The boring case is where a phoneme has only one allophone (pretty rare, actually). "Near minimal pair" is about the phonetic record, so it is always a relation between phones i.e. allophones. The reason for confusion over what a minimal pair is, is that this kind of automated distributional analysis has fallen out of favor, even though we retain the old terminology.