Different /ə/ pronunciation at the end of a word; for example, in "phenomena"

Solution 1:

On /phonemic/ vs [phonetic] Transcriptions

What’s going on here is that some dictionaries are sometimes using detailed phonetics yet placing them within the slashes that should be used only for phonemic transcriptions, not for phonetic ones.

Remember that everything in slashes always represents a range of possible phonetic allophones that can vary without the meaning of the word ever changing in native speakers’ minds. Phonemes are abstract things; they’re what people think they hear being said. Phonetics are something else, the actual sounds being produced.

So some people may well say [fəˈnɒmɨnɐ] one day, while other people or even the same people on a different day may say [fəˈnɑmɪnə]. But no native speaker will ever think that those are two different words. That’s because the degree of vowel reduction in unstressed syllables is not a phonemic distinction, and there is no such thing as an /ɐ/ phoneme in English (unlike for example in European Portuguese).

Properly written as a broad phonemic transcription, that should probably just be phonemic /fəˈnɒmənə/. However, this requires that the reader be well-versed in English phonology. Otherwise they won’t realize that:

• The /ə/ phonemes in that word can variously be expressed as [ə], [ɪ], [ɪ̈], [ɨ], or [ᵻ] — or even [ɐ] as you have yourself noticed.
• The /ɒ/ phoneme can similarly be [ɔ], [ɒ], or [ɑ].

So you aren’t supposed to use slashes there, but dictionaries are prone to doing so. But you can’t take them seriously: for example, you can immediately tell that they’re actually trying to indicate phonemics because they claim that right and write are pronounced /raɪt/, but as phonetics that’s wrong for just about anyone outside of Scotland (nobody else has a trilled/rolled r). Phonetically, right and write are often [ɻʌɪt̚] for many speakers, but hardly for all.

Learning which phonetic allophones should all go into each phonemic bucket is one of the hardest things for non-native speakers to get a handle on. That’s because until they do, they won’t know whether a different pronunciation actually means a different word or not. If it does, it’s a phonemic distinction, but if it does not, then it’s merely a phonetic one.

For actual phonetics, the UK Sound Comparisons website is incomparable. See for example all the different phonetics for the word naked there.

Solution 2:

In some accents of English (I think it's particularly common in American English specifically), word-final "/ə/" is pronounced with an opener quality than word-internal "/ə/". Word-final /ə/ may be something like [ɐ], [ɜ] or [ʌ] (for me, it sounds a lot like the stressed vowel of "cut"), while word-internal /ə/ is usually a fairly close vowel—in the comments, John Lawler suggested that [ɨ] would be an appropriate transcription. I would guess that it's relevant that many American English speakers have the "weak vowel merger", where the distinction between /ɪ/ and /ə/ in fully unstressed syllables is mostly neutralized: the neutralized vowel phoneme is conventionally transcribed /ə/, but its phonetic realization is fairly variable.

Note: although I said "word final", for me and many other speakers the opener quality is also used in affixed forms like Sarah's, subpoenaed. The potential contrast between originally word-final schwa and another weak vowel is often described in terms of the minimal pair "Rosa's" vs. "roses".

The difference in quality between "word final" /ə/ and word-internal /ə/ is usually not transcribed because it's considered an allophonic detail rather than a phonemic contrast between /ə/ and /ɐ/.

I wrote some more about conditioned allophonic variation in the realization of the "schwa" phoneme in my answers to the following questions:

• Exodus word pronounciation - Why it is different from spelling

• What exactly is the “schwa” sound?