Appendix talk:Egyptian pronunciation

Just passing by to say that, even though I'm not interested in Egyptian, I find this beautiful. --Per utramque cavernam (talk) 19:08, 27 March 2018 (UTC)

J.P.Allen has a new book "Ancient Egyptian Phonology" (→ISBN, https://doi.org/10.1017/9781108751827) which should be added to your future references. ODConstant (talk) 11:08, 14 June 2021 (UTC)

@ODConstant: Yep, I’m aware of it (and have read it, actually!), but it’s been a while since I worked on this page and would need to go through all my sources again to get back up to speed on what other authors say… Beyond that, though, Allen doesn’t really go deeply into the evidence like Peust did; he gives something more along the lines of a summary of his understanding, which in some cases doesn’t align with the previous academic consensus. As of now, my understanding is that some of these conclusions have yet to become widely accepted (especially in Germany, where most of the research in Egyptian phonology has historically taken place). We’ll see where future research takes us, but I think for now the best stance to take is wait-and-see. — Vorziblix (talk · contribs) 12:44, 14 June 2021 (UTC)

Some issues with Egyptian pronunciation on Wiktionary

Hi @Vorziblix. I’ve been noticing lately that some of the Old/Middle Egyptian reconstructed pronunciations I see on Wiktionary go far beyond what most scholars in the field today would be confident reconstructing. I don't mean this with regards to you, but I think a lot of editors who are only somewhat familiar with Egyptian see this page and are left with the impression that there’s a greater degree of certainty to this stuff than there actually is. There's a lot of great content in this appendix, but I think it needs to be clearer about the state of our knowledge. Many of the phonetic values for these consonants in Old/Middle Egyptian and the specifics of some of these sound changes are just not known, or are otherwise heavily debated. On the other hand, certain things that are known aren't covered here. It seems like most editors working on Egyptian entries use the material here as the basis for their reconstructions, but little of what they produce is in agreement with either Osing (1976) or Vycichl (1983), which are the two major sources for reconstructed Egyptian words. Neither is it really reflective of the current state of the field.

Here are some preliminary notes on what I think could be improved regarding the symbols used:

• The existence of ejective consonants in Egyptian is speculative. On the other hand, that p t ṯ k were aspirates is a known fact directly attested by Coptic. It’s only responsible that we represent the distinction between the p t ṯ k and d ḏ g q series using the actually attested feature rather than the speculative one. This is the approach taken by the two most recent major studies of Egyptian phonology (Peust 1999 and Allen 2020).
• We are simply not yet capable of identifying the features that originally distinguished q from g, ẖ from ḫ, and z from s. There is a complete lack of consensus in the field regarding their phonetic values. In reconstructions we should distinguish them by using some sort of neutral cover symbol like ⟨k^◌ x^◌⟩ for q and ẖ, and ⟨S⟩ or ⟨Z⟩ for z.
• The value /ʀ/ for ꜣ is largely idiosyncratic to Loprieno and has not been supported by subsequent studies. For obvious reasons, it’s also highly speculative. Note that both Peust and Allen now support a value of /l/ for ꜣ. In his recent review of Allen’s Ancient Egyptian Phonology, Peust stated quite directly that:

"Für den Lautwert in früherer Zeit würden zwar weitere Möglichkeiten offenbleiben, sofern man zusätzliche Lautwandel ansetzt, aber die stärkste verfügbare Evidenz spricht nun einmal für die auch von Allen bevorzugte Interpretation ⟨ꜣ⟩ = /l/. Diese sollte man künftig übernehmen, solange nicht stärkere Gegenargumente auftauchen."

I’m sympathetic to Peust and Allen’s view because the value /l/ is directly attested for Coptic reflexes of ꜣ, but Wiktionary could just as well use a cover symbol like ⟨R⟩.

• For r, there’s no need to use ⟨ɾ⟩ rather than ⟨r⟩. No other source does this, and it's generally a good idea to keep phonological transcriptions of dead languages broad.
• /ʔ/ should probably be marked with parenthesis or made optional. The position that there was a phonemic glottal stop in Egyptian is increasingly marginalized and has been rejected by both Peust and Allen.
• In foreign transcriptions going as far back as there is analyzable data, ṯ and ḏ are exclusively equated with Semitic and Anatolian affricate consonants. The Coptic reflexes ϫ (and Bohairic ϭ) are likewise unambiguously affricates. We should represent these phonemes accordingly as /t͡ʃʰ t͡ʃ/ rather than as stops.

For Middle Egyptian, a more even-handed synthesis of the present state of the field might look like:

Middle Egyptian consonants

		Labial	Alveolar		Palatal	Velar		Pharyngeal	Glottal
Nasal		m ⟨m⟩	n ⟨n⟩
Obstruent	aspirate	pʰ ⟨p⟩	tʰ ⟨t⟩		t͡ʃʰ ⟨ṯ⟩	kʰ ⟨k⟩			(ʔ)
	tenuis	b ⟨b⟩	t ⟨d⟩		t͡ʃ ⟨ḏ⟩	k ⟨g⟩	k◌ ⟨q⟩
	fricative	f ⟨f⟩	s ⟨s z⟩		ʃ ⟨š⟩	x ⟨ḫ⟩	x◌ ⟨ẖ⟩	ħ ⟨ḥ⟩	h ⟨h⟩
Approximant			r ⟨r⟩	l ⟨ꜣ⟩	j ⟨j⟩	w ⟨w⟩		ʕ ⟨ꜥ⟩

1. Where ⟨^◌⟩ stands for an indeterminate secondary articulation.

Regarding the sound change rules presented here, the relative chronology is inaccurate in some places. Schenkel’s Law is contradicted for instance, for which, see Peust’s Review of Allen:

"Anachronistisch sind meiner Auffassung nach des Weiteren solche Ansätze Allens wie /w˘-naˈ-wa/ für wnw.t „Stunde“ (S. 44) und /saˈ-wa/ für swr „trinken“ (S. 55), denn der Verlust des -w- durch das Schenkelsche Gesetz „Intervokalisches w nach dem Tonvokal wird zu j“ trat schon in der Ersten Zwischenzeit ein und damit ebenfalls vor dem Abfall des silbenschließenden ⟨t⟩ und ⟨r⟩." (Edit: actually this is correct in the appendix, just not followed in some entries)

Other rules such as Edel’s law of finales need to be added. Vowels are a discussion for another day, but I'd also like to note that the terms and dates used by Template:egy-IPA-R for the stages of Egyptian are also kind of idiosyncratic. Going off the definitions given there, “Medio-Late Egyptian” should be changed to Late Egyptian and “Late Egyptian” to Demotic. One simple thing that could be done to improve the quality of reconstructions is to explicitly introduce a rule against reconstructing words without Coptic reflexes or suitable foreign transcriptions. It’s obvious, but a lot of entries currently violate this. Rhemmiel (talk) 13:20, 30 July 2021 (UTC)

@Rhemmiel Hi! Yes, I agree there is a great deal of uncertainty in this field, and some of our reconstructions go beyond what we can be confident about here. To be honest, a lot of the mess in our Egyptian pronunciations (and also the incompleteness of this appendix, from which some of that mess derives) is in part due to real-life causes: I had a change of life circumstances in the middle of sorting all of this out back in 2018 and haven’t had enough time to work on fixing/finishing it since then. I was in the process of sourcing every reconstructed pronunciation that existed to that date, adding new ones from Loprieno and other sources, getting rid of those that I couldn’t source, and adding more missing material to this appendix, but the project was never finished. Since then other editors have stepped in and added copious reconstructions of their own, which unfortunately I have had not always had the time or the expertise to fully verify.

To some of your more specific points:

• I think that, in the paragraphs in the top of this appendix, the uncertainty you talk about is directly acknowledged: ‘a possible sequence of sound changes’, ‘Details are generally much less clear for earlier periods’, ‘details remain unclear’, etc. I’m not really sure how we would make it clearer about the state of our knowledge, as I did try to make it pretty explicit, but if you have any suggestions, we could certainly add even more qualifiers.
• The first section also explicitly mentions that, as you say, there is no agreement about what distinguished ⟨q⟩ from ⟨g⟩ or what values ⟨z⟩ and ⟨ꜣ⟩ had. In this case the symbols we use are purely conventional ones, but they were broadly chosen to agree with Loprieno’s notation. I wouldn’t object to using some other symbols, but I don’t really see what advantage using, say, /k^◌/ or /S/ would confer; /q/ and /z/ are also intended as mere cover symbols, and the page explains that, but at least they’re ones that other publications have also used.
• I disagree regarding ⟨ẖ⟩ and ⟨ḫ⟩; the arguments for non-traditional values for these two sounds are almost exclusively based on arguments for certain Afro-Asiatic sound correspondences, which are IMO extremely flimsy (see the enormous gap between the proposals of the neuere Komparatistik folks and those of traditionalists like Takács, with no clear consensus to be found among academics as a whole).
• I wouldn’t object to changing /ʀ/ to something else, including /l/. However, in that case, we must decide what symbol to use to represent the different /l/ sound that appears in ns (“tongue”) and similar words.
• I also wouldn’t object to changing /c/ to /t͡ʃ/, and agree with your analysis. In fact, we do use /t͡ʃ/ for Middle Egyptian and everything later, for exactly the reasons you mention; it’s only for Old Egyptian that we have /c/, and this was basically for two reasons: (1) as a concession to the common use of /c/ as a broad transcription in the sources, without intending to imply, however, that that was what the actual narrow realization of that sound might be; and (2) as an acknowledgement that the sound probably descends from a palatalized /k/, which makes it likely that it first passed through a /c/ stage on its way to becoming /t͡ʃ/. Even conceding that point (2) is likely, however, there’s really no reason to assume the change to /t͡ʃ/ happened after what we label as ‘Old Egyptian’.
• A value of [ɾ] for ⟨r⟩ seems pretty solidly established and widely acknowledged, based on evidence like the transcription of Semitic /d/ as ⟨r⟩, which would be quite unlikely if ⟨r⟩ was a proper trill but perfectly ordinary for a tap. Practically every paper and study I’ve seen that discusses one realization versus the other concludes that ⟨r⟩ was tapped, from Hoch 1994 to Loprieno 1995 to Allen 2013. (In his recent Ancient Egyptian Phonology, Allen stays ambiguous by saying tap/trill and giving both realizations, but only provides evidence in favor of the tap. It’s unclear to me why he does this. Unfortunately he provides no further justification. Edit: Actually, later in the study, he explicitly says /r/ was ‘clearly’ [ɾ] ‘at least from Middle Egyptian onward’.) While I would support a broad transcription in cases where things are uncertain, this seems to be one of the few points of phonology where almost everyone does agree. Using /r/ for [ɾ] in those circumstances seems a bit like using /r/ for English [ɹ] — it makes sense for convenience’s sake in a publication discussing a single language, but IMO it’s more confusing than helpful in a multilingual dictionary.
• I have no strong opinion on the glottal stop. If we do decide to get rid of it, we’ll have to decide what to do about certain words following Loprieno’s reconstruction such as nwt (“city”).
• The aspirated vs. glottalized/emphatic/ejective debate is a giant can of worms. Peust is agnostic on the question, Allen (as you say) decides in favor of the aspirates, while most of the Germans together with Loprieno take the side of the emphatics. Aspiration is indeed attested in Coptic, but the distribution is not the same, and we can’t decisively conclude from it whether earlier ⟨p t ṯ k⟩ were in fact aspirates or just had aspirated allophones (as with English tenuis stops). Given this, I wouldn’t say that either the equations of ⟨p t ṯ k⟩ = /pʰ tʰ t͡ʃʰ kʰ/ or that of ⟨b d ḏ g⟩ = /b d’ t͡ʃ’ k’/ as a series are directly attested. I think the choice between them as representations is essentially arbitrary.

Now, this is just speculation, but personally I think the distinction was one of both glottalization (ejectives/implosives) and aspiration, and that the system of Egyptian plosives was basically like the one we find in Kʼicheʼ Mayan. This has the advantage of perfectly explaining the anomalous voiced /b/, which would otherwise be the only ordinary voiced plosive: if all the tenuis stops were glottalized, an implosive articulation (which is usually voiced) is typologically most likely for labial sounds, while other glottalized sounds are more likely to surface as ejectives (which are never voiced).

• I haven’t read Peust’s review of Allen, but would be very interested in doing so and grateful for a pointer in that direction. Would you happen to have a link, by any chance? (Similarly, I’ve been trying to get my hands on Osing 1976 for years with no luck… if by chance you know of any place to read it online, or even to buy it in physical form for anything less than a fortune, I’d be grateful.)
• Regarding terminology, ‘Medio-Late Egyptian’ is a term used by Friedrich Junge and others for the colloquial language of precisely the period marked as such in the template (Amarna). I’m also not aware of any authors extending Demotic back to 800 BCE. However, I concede that I could very well be wrong about that, as I’m far from an expert on Early Demotic.

I’m aware that things are missing from this page, and that some changes might need to be re-arranged. I know that especially a lot of the vowel changes from Late Egyptian through Demotic are missing, because unfortunately I never got around to adding them. The page, as it is, remains a work in progress. Feel free to add anything you find isn’t there or fix anything you find that’s obviously wrong.

Your proposed general rule as to what reconstructions to allow also seems like a good one. Thanks for taking the time to go through all this! — Vorziblix (talk · contribs) 00:24, 31 July 2021 (UTC)

@Rhemmiel Any thoughts? I’d really like to get this cleaned up and more in line with current thinking, and it’s clear you have a depth of knowledge in the field. (Not to rush you if you need more time to formulate a response.) Also pinging @BoxGirder for a third opinion. — Vorziblix (talk · contribs) 05:49, 11 August 2021 (UTC)

Hi @Vorziblix and @Rhemmiel! I agree with most of the points mentioned above. Updating the contents of this appendix and fixing the messy and (sometimes) questionable reconstructions in some entries sound good to me.

Thoughts:

• Given the long history of the Egyptian language, it's understandable that there are various uncertainties regarding the actual values of some phonemes. However, I think it's acceptable to use values based on relevant studies (even if those values are still debated), as long as the uncertainties and alternatives are stated.
• I don't have strong opinions regarding the distinction of Egyptian plosives. I think all the mentioned interpretations are possible, but I'm currently more inclined towards the aspirable/aspirated-plain distinction (primarily because, as mentioned above, ejectives are never voiced, but ⟨ⲧ ϫ⟩ became voiced around 1300 CE). Either way, we should still note the various interpretations before the large table.

Regarding the aspirated-glottalic distinction, I think that's possible for early Egyptian. The glottalization was then dropped at some point, and the resulting plain voiced /b/ became spirantized in Late Egyptian, while the plain voiceless /t t͡ʃ/ became voiced in Late Coptic.

• The Egyptian /l/ phoneme is a tricky one, but it's possible that the /l/ of ns is the same /l/ written using ⟨ꜣ⟩. It's unclear why Old Egyptian /l/ was sometimes written as ⟨n⟩, but I guess it's probably because /n/ and /l/ are related (based on Afroasiatic cognates), and because ⟨ꜣ⟩ started to lose its consonantal value earlier in the dialect of Old Egyptian than in the dialect of Middle Egyptian.

As for reconstructions, we could use /l/ for all instances of earlier Egyptian ⟨ꜣ⟩ and for ⟨n⟩ that becomes Coptic ⟨ⲗ⟩. On the other hand, both /ɾ/ and /l/ could be used for ⟨r⟩ that becomes ⟨ⲗ⟩, since there's evidence for a sound change [ɾ] → [l] before Demotic: e.g. ḫꜣrw (Middle Babylonian ḫu-ra → Neo-Assyrian ḫi-ri) = ẖl = ϩⲁⲗ.

• Egyptian ⟨r⟩ is a tap [ɾ]. Using /r/ in Wikipedia and Wiktionary can be misleading, especially since the IPA template links the symbol to [r] rather than [ɾ].

Also, [r] is not an approximant. Readers may confuse an approximant /r/ with [ɹ] (alveolar approximant), which is different from the typical taps and trills of Afroasiatic languages.

• Old Egyptian ⟨z⟩ was probably voiceless like the other fricatives, so I guess using /z/ as a cover symbol can also be confusing? We could use a different value, but we still have to mention the uncertainty before the large table (same with /l/).
• Old Egyptian ⟨š⟩ and Middle Egyptian ⟨ẖ⟩ both represented a palatalized dorsal fricative. This phoneme is usually interpreted as palatal /ç/, but it's probably a palatalized velar /xʲ/.
• The difference between velars ⟨g q⟩ is indeed not evident. Their Coptic reflexes are contrasted by palatalization, so I'm currently inclined towards the values /kʲ k/ for ⟨g q⟩.
• I wouldn't object to dropping the glottal stop. However, the related ⟨j⟩ is a bit complicated. Since ⟨j⟩ sort of worked as a mater lectionis during the Middle (and possibly Old) Kingdom, I think it's possible that the ⟨j⟩ in some Old Egyptian words actually represented a vowel rather than /j/ (e.g. jnk and jzr, based on Afroasiatic cognates).

As for reconstructions, we could drop all instances of /ʔ/, except for those that result to Coptic doubled vowels, which could either be dropped or be marked with parentheses or brackets. (We could probably just replace the /ʔ/ in nwt with /w/.)

• ⟨ṯ ḏ⟩ are the palatalized counterparts of ⟨t d⟩. Affricates /t͡ʃʰ t͡ʃ/ are palato-alveolars, so I also wouldn't object to using those values even for Old Egyptian.
• I think the parameters' labels are just right: the Neo-Assyrian inscriptions indeed fall within Demotic, but most of the changes reflected in those inscriptions already happened during the Third Intermediate Period, which still falls within Late Egyptian.
• The proposed rule will clean up the messy reconstructions in a bunch of entries, so it also sounds good to me.

I also made some consonant tables. These are similar to the one above, but I made some modifications.

Old Egyptian consonant phonemes

		bilabial	dental	alveolar	postalveolar	palatal	velar		pharyngeal	glottal
							palatalized	plain
nasal		m ⟨m⟩		n ⟨n⟩
plosive, affricate	aspirated	pʰ ⟨p⟩		tʰ ⟨t⟩	t͡ʃʰ ⟨ṯ⟩			kʰ ⟨k⟩
	plain			t ⟨d⟩	t͡ʃ ⟨ḏ⟩		kʲ ⟨g⟩	k ⟨q⟩
	voiced	b ⟨b⟩
fricative		ɸ ⟨f⟩	θ ⟨z⟩	s ⟨s⟩			xʲ ⟨š⟩	x ⟨ḫ⟩	ħ ⟨ḥ⟩	h ⟨h⟩
approximant		w ⟨w⟩		l ⟨ꜣ⟩		j ⟨j⟩			ʕ ⟨ꜥ⟩
tap				ɾ ⟨r⟩

Middle Egyptian consonant phonemes

		bilabial	alveolar	postalveolar	palatal	velar		pharyngeal	glottal
						palatalized	plain
nasal		m ⟨m⟩	n ⟨n⟩
plosive, affricate	aspirated	pʰ ⟨p⟩	tʰ ⟨t⟩	t͡ʃʰ ⟨ṯ⟩			kʰ ⟨k⟩
	plain		t ⟨d⟩	t͡ʃ ⟨ḏ⟩		kʲ ⟨g⟩	k ⟨q⟩
	voiced	b ⟨b⟩
fricative		ɸ ⟨f⟩	s ⟨z s⟩	ʃ ⟨š⟩		xʲ ⟨ẖ⟩	x ⟨ḫ⟩	ħ ⟨ḥ⟩	h ⟨h⟩
approximant		w ⟨w⟩	l ⟨ꜣ⟩		j ⟨y⟩			ʕ ⟨ꜥ⟩
tap			ɾ ⟨r⟩

Late Egyptian and Demotic consonant phonemes

		bilabial	alveolar	postalveolar	palatal	velar		pharyngeal	glottal
						palatalized	plain
nasal		m ⟨m⟩	n ⟨n⟩
plosive, affricate	aspirated	pʰ ⟨p⟩	tʰ ⟨t⟩	t͡ʃʰ ⟨ṯ⟩			kʰ ⟨k⟩
	plain		t ⟨d⟩	t͡ʃ ⟨ḏ⟩		kʲ ⟨g⟩	k ⟨q⟩
fricative	voiceless	ɸ ⟨f⟩	s ⟨z s⟩	ʃ ⟨š⟩		xʲ ⟨h̭ ḫ⟩	x ⟨ẖ⟩	ħ ⟨ḥ⟩	h ⟨h⟩
	voiced	β ⟨b⟩
approximant		w ⟨w⟩	l ⟨l⟩		j ⟨y⟩			ʕ ⟨ꜥ⟩
tap			ɾ ⟨r⟩

Coptic consonant phonemes

		bilabial	alveolar	postalveolar	palatal	velar		glottal
						palatalized	plain
nasal		m ⟨ⲙ⟩	n ⟨ⲛ⟩
plosive, affricate	aspirated	pʰ ⟨ⲫ⟩	tʰ ⟨ⲑ⟩	t͡ʃʰ ⟨ϭ⟩			kʰ ⟨ⲭ⟩
	plain		t ⟨ⲧ⟩	t͡ʃ ⟨ϫ⟩		kʲ ⟨ϭ⟩	k ⟨ⲕ⟩
fricative	voiceless	ɸ ⟨ϥ⟩	s ⟨ⲥ⟩	ʃ ⟨ϣ⟩		xʲ ⟨ⳃ ⳋ⟩	x ⟨ϧ ⳉ⟩	h ⟨ϩ⟩
	voiced	β ⟨ⲃ⟩
approximant		w ⟨ⲟⲩ⟩	l ⟨ⲗ⟩		j ⟨ⲉⲓ⟩
tap			ɾ ⟨ⲣ⟩

Late Coptic consonant phonemes

		labial	alveolar	postalveolar	palatal	velar	glottal
nasal		m ⟨ⲙ⟩	n ⟨ⲛ⟩
plosive, affricate	voiceless		t ⟨ⲑ⟩			k ⟨ⲕ ⲭ⟩
	voiced	b ⟨ⲡ⟩	d ⟨ⲧ⟩	d͡ʒ ⟨ϫ⟩
fricative		f ⟨ϥ⟩	s ⟨ⲥ⟩	ʃ ⟨ϣ ϭ⟩		x ⟨ϧ⟩	h ⟨ϩ⟩
approximant		w ⟨ⲃ ⲟⲩ⟩	l ⟨ⲗ⟩		j ⟨ⲓ⟩
tap			ɾ ⟨ⲣ⟩

Notes and modifications:

• I rearranged the consonants according to their positions in the IPA chart.
  • /j/ is isolated to form an axis of symmetry between the palato-alveolars and the palatalized velars.
  • /b β/ are isolated to highlight their anomalous voicing.
  • /w/ is placed in the bilabial column to reflect its alternation with /β/ in Coptic.
• /ɸ β/ can also be /f v/. I presented them here as bilabials to reflect their alternation and relationship with earlier ⟨p b⟩.
• Consonants during the Amarna period are similar to Middle Egyptian's, but with /l/ written using ⟨n r⟩ instead of ⟨ꜣ⟩.
• Consonantal values are based on Allen (2020), except for the affricates, earlier Egyptian ⟨f⟩, Old Egyptian ⟨j⟩, and Coptic ⟨ϩ⟩.

Regarding vowels, there are some conflicting interpretations that need to be addressed as well (e.g. vowel length vs. quality, Coptic doubled vowels, vocalic patterns, epenthetic vowels and consonant clusters).

Anyway, sorry for the late reply! — BoxGirder (talk) 11:09, 1 September 2021 (UTC)