Wiktionary says that "grass" is supposed to be pronounced /ɡɹ̠æs/ in General American English (https://en.wiktionary.org/wiki/grass), but then the US audio sample clearly pronounces it with a long /æ/.
I think it's because of this (https://en.wikipedia.org/wiki/General_American_English):
Vowel length is not phonemic in General American, and therefore vowels such as /i/ are usually transcribed without the length mark. Phonetically, the vowels of GA are short [ɪ, i, ʊ, u, eɪ, oʊ, ɛ, ʌ, ɔ, æ, ɑ, aɪ, ɔɪ, aʊ] when they precede the fortis consonants /p, t, k, tʃ, f, θ, s, ʃ/ within the same syllable and long [ɪː, iː, ʊː, uː, eːɪ, oːʊ, ɛː, ʌː, ɔː, æː, ɑː, aːɪ, ɔːɪ, aːʊ] elsewhere. This applies to all vowels but the schwa /ə/ (which is typically very short [ə̆]), so when e.g. /i/ is realized as a diphthong [i̞i] it has the same allophones as the other diphthongs, whereas the sequence /ɜr/ (which corresponds to the NURSE vowel /ɜː/ in RP) has the same allophones as phonemic monophthongs: short [ɚ] before fortis consonants and long [ɚː] elsewhere. The short [ɚ] is also used for the sequence /ər/ (the LETTER vowel). All unstressed vowels are also shorter than the stressed ones, and the more unstressed syllables follow a stressed one, the shorter it is, so that /i/ in lead is noticeably longer than in leadership.
Both Wiktionary and Cambridge English Dictionary also say that the word "ass" is pronounced as /æs/ in both American English and British English. But at least in US English, "ass" is more often pronounced like /æːs/ and "as" is pronounced like /æz/, even though for example the word "big-ass" can be pronounced like /bɪɡæs/.
Therefore at least in US English, the words "as" and "ass" don't follow the pattern described in the quotation above, because in the word "as" (/æz/), /z/ is a lenis consonant but it is preceded by a short vowel, and in the word "ass" (/æːs/), /s/ is a fortis consonant but it is preceded by a long vowel.
Also in some American English accents that have undergone the "cot-caught merger", there are minimal pairs that demonstrate a non-allophonic distinction between /ɑ:/ and /ɑ/, like for example "caught" (/kʰɑ:t/) vs "cut" (/kʰɑt/) and "stalk" (/stɑ:k/) vs "stock" (/stɑk/) .
I don't agree with this statement (https://en.wikipedia.org/wiki/Cot–caught_merger): "The shift causes the vowel sound in words like cot, nod and stock and the vowel sound in words like caught, gnawed and stalk to merge into a single phoneme; therefore the pairs cot and caught, stock and stalk, nod and gnawed become perfect homophones, and shock and talk, for example, become perfect rhymes."
Even after the "cot-caught merger", there is still a clear distinction in vowel length between words like "nod" (/nɑd/) and gnawed (/nɑ:d/).
English orthography is just so messed up that English is written non-phonetically even in IPA. Just because distiction in vowel length in omitted from IPA transcription doesn't mean it doesn't exist.
Bookmarks