Module talk:ha-IPA

@Malku_H₂n̥rés: Seeing as the narrow Kano pronunciation is currently incorrect, here are my notes from Newman's chapter on phonology in his encyclopaedia. Note that palatalisation and labialisation occur before any of the vowel sound changes.

1. Considerable idiolectal variation in /f/. I think that [φ] is acceptable from the perspective of Standard Hausa, but less standardly in Kano one finds [p] before unrounded vowels and [h] before /o/ and /u/. Perhaps a reason to mark this as standard rather than Kano?
2. Newman (and my experience) point to using just [sʼ] for <ts>, but we could include (t) before that if you want.
3. WH dialects can have labialised coronals and labials, and these should be added to the pandialectal broad transcription for completeness's sake. The sake is true of <c'>.
4. All dialects have word-final /n/, as well as /n/ immediately before a velar, as [ŋ]. Standard/Kano also has final /m/ as [ŋ], but not in ideophones, so an override is needed.
5. All dialects have coda /n/ assimilate to the following consonant (to [ŋ] before velars and glottals, to [m] before bilabials and /w/, and to [ɲ] before palatals). Before the liquids (<l>, <r>, <r̃>), /n/ assimilates to match them (so /n.l/ becomes [l.l]). All assimilation occurs across word boundaries, so it needs to ignore a space or hyphen.
6. In Kano, /r/ tends to assimilate to a following coronal. I think this is another one we can potentially ignore if we call it standard instead.
7. The vowels are exceedingly messy. We may want to refine this with other sources, but here is a straightforward set of rules we can implement. Note that short vowels always include those in closed syllables.
   1. Long vowels are to match the phonemic notation.
   2. Short /a/ to be [ɛ] when followed by <y> or <ny>, or preceded by <y> or <ƴ> (therefore including <'y> and palatalised consonants), [a] in other closed syllables or if preceded by [ʔ] or [h], and [ə] otherwise.
   3. Short /o/ preceded by /h/ to be [ɔ] in closed syllables only. Short /o/ and /e/ to be [a] in all other closed syllables, and [o] or [e] (respectively) otherwise. In recent loanwords they can be preserved as [ɔ] or [ɛ] (respectively) in closed syllables; we probably want an override for this.
   4. Short /i/ to be [ɪ].
   5. Short /u/ before /j/ to be [i], following a coronal to be [ɪ], and [ʊ] otherwise.
   6. The diphthong /ai/ to be [ai] following /ʔ/ or /h/, or in a monosyllabic word with falling tone, and [ei] elsewhere.
   7. The diphthong /au/ to be [au]. (Really, it's all over the place, and Newman seems to default on [ao], but I hear something more like [əu].)
8. An important consideration is whether we want to represent the word in pausa. In prepausal position, all short vowels are followed by [ʔ], and a subset of long vowels (determined morphologically and lexically) are shortened (and considered half-long, although it's not clear that they can be distinguished from short vowels) and followed by [ʔ]. This would be a lot of extra work to implement, so I would honestly prefer to ignore it!

—Μετάknowledge^{discuss/deeds} 00:10, 15 July 2021 (UTC)

@Metaknowledge: some questions:

1. Considerable idiolectal variation in /f/. I think that [ɸ] is acceptable from the perspective of Standard Hausa, but less standardly in Kano one finds [p] before unrounded vowels and [h] before /o/ and /u/. Perhaps a reason to mark this as standard rather than Kano?
   1. Should I include it? This means that what's displayed as /f/ becomes /ɸ/, so that [ɸ] is replaced by [p] and [h]. Honestly, would it be a good idea to separate a (Kananci-based) Standard from the Kano speech, looked upon as a dialect? And isn't it [hʷ]? Eventually, you wrote /o/, this means that it's [h] even when /o/ is [a]?
2. WH dialects can have labialised coronals and labials, and these should be added to the pandialectal broad transcription for completeness's sake. The sake is true of /c'/.
   1. It applies only to Western Hausa dialects, but I'll follow you and do this, however you just said they can occur, not the conditions making them happening. It applies to alveolars, post-alveolars, liquids, and the ejective voiceless palatal stop, right?
3. Note that short vowels always include those in closed syllables.
   1. Effectively in testcases, closed syllables always have a short vowel, conversely all long vowels are in open syllables (excepted Allā̀h). Is it perfectly regular otherwise?
4. An important consideration is whether we want to represent the word in pausa. In prepausal position, all short vowels are followed by [ʔ], and a subset of long vowels (determined morphologically and lexically) are shortened (and considered half-long, although it's not clear that they can be distinguished from short vowels) and followed by [ʔ].
   1. If it's not impossible to do and it's quite useful, I can do this. I just need the clear rules to manage it.

Otherwise I assumed I shouldn't modify anything, such as /au/ as [au], when it's unclear. Malku H₂n̥rés (talk) 08:53, 16 July 2021 (UTC)

Before I begin, I need to ensure the phonetic pronunciations at Module:ha-IPA/sandbox/testcases are correct.

@Malku H₂n̥rés: 1. This is a sticky issue; I'd prefer just to stick with [ɸ]. 2. Not all of those, no. I don't know the exhaustive list present in Hausa dialects, unfortunately, but I'm pretty sure there are gaps. 3. Yes, and Allah#Hausa gets respelt anyway. 4. It's not impossible, but the number of special cases that would have to be added would seriously complicate automation, so I'd like to avoid it, at least for now. —Μετάknowledge^{discuss/deeds} 04:36, 19 July 2021 (UTC)