Talk:Palatalization in the Romance languages

IPA
Not sure if it's just my computer, but the IPA in this article is not showing up for me when I view the article, and is only visible as a template when I attempt to edit the article. They show up fine in other articles.

Pinging Urszag  as they've been working on this article and I'm here to review it as a new page. -- NotC hariza rd 🗨 10:17, 15 September 2023 (UTC)


 * Okay nevermind it seems to be working fine now. Maybe my computer just blipped out! -- NotC hariza rd  🗨 10:18, 15 September 2023 (UTC)

'Learned'
That French reflexes of the type étrange, linge, lange< extrāneum, līneum, lāneum would be 'learned' is a bit difficult to imagine. There is nothing semantically learned about them, and they are otherwise regular in their sound-changes. In light of the fact that /nj/ > /nd͡ʒ/ is parallel to the more regular /mj/ > /nd͡ʒ/ (cf. singe, vendange < sīmium, vindēmia), it is not clear why the former should need a special explanation at all.

To the best of my knowledge, none of the widely-cited reference works (Von Wartbug's FEW, Pope's From Latin to French, and so on) claim such words to be learned. Considering that the only citation provided for this notion is Brittain 1900, perhaps it would make sense to preface it with a caveat or remove it. Nicodene (talk) 05:37, 21 September 2023 (UTC)
 * Thanks for bringing this up, I was not sure about including it. I think I saw this explanation in another source, which I will try to find. It appears to be less common than /ɲ/, at any rate, and I see no phonological conditioning factor, so it calls for some explanation, even if this one might not be certain. I also saw another explanation, which I find hardly convincing but might be worth mentioning, which explained it as the result of analogical restoration of alveolar/dental [n] from related forms or by-forms.--Urszag (talk) 06:34, 21 September 2023 (UTC)
 * OK, I just saw that Kerkhoff 2018 is another source that describes /nd͡ʒ/ as a learned outcome (specifically, "linge" and "lange" are described as borrowings from Latin into Old French). This is cited to Pope 1934: 229 and Zink 1986: 228-29, so I will follow up on those when I have the chance.--Urszag (talk) 07:22, 21 September 2023 (UTC)
 * I do not have access to Zink (1986) but can provide the quote from Pope (1934: 229):
 * 'In Gallo-Roman, particularly in Early Gallo-Roman, the clerks carried over into their pronunciation of Latin the tendencies that were most strongly affecting the speech of the time, and the words borrowed at this epoch ordinarily reflect these developments. The most marked are the following: [...] The consonantisation of e and i unstressed in hiatus before a vowel and the passage of j to dz̆ when retained as initial (§§220, 203): O.F. mili̯e < millia, pali̯e < pallium, linge < linea, lange < lanea, serorge < sororium.'
 * I stand corrected; apparently Pope did consider the /nj/ > /nd͡ʒ/ outcomes to reflect early borrowing. The FEW however speaks of linge/lange/étrange as inherited terms, as does the TLFi. I follow the latter two and disagree with Pope on the grounds mentioned earlier, particularly the parallel with /mj/ > /nd͡ʒ/ (cf. also the affricated Sardinian outcome, where learned influence is even less likely). It would also be unclear why pali̯e < pallium, which is indeed a learned/ecclesiastical term, did not develope an affricate if one follows Pope's reasoning.
 * The assumption that, when there are two outcomes of one original sequence, one of them must be an aberration in need of a special explanation, is based on the old neo-grammarian hypothesis of exceptionless sound-change, which runs aground on the actual data. We see multiple instances on this very page; it would be a brave man who labels the Spanish sepa < sapiat as learned in light of jibia < sepia.
 * Still, that Pope, etc. believed so does bear mention. All I have to add, regarding your most recent formulation, is that 'learned forms (borrowed into Old French from Latin)' and 'introduced later to the language before undergoing popular development' are describing the same thing. Nicodene (talk) 18:12, 21 September 2023 (UTC)
 * Yes, I was still working on the wording. I have reformulated it now to something I think makes more sense, although maybe I should just leave out the word "learned" rather than giving a disclaimer about how it doesn't mean the same thing in this case as "savant" in "mot savant". I couldn't figure out what chronology Pope had in mind when citing mili̯e and pali̯e alongside linge and lange as examples of borrowed /j/. I don't think anyone alleges learned influence in the case of Sardinian, but there the affricated outcome is regular (as far as I know) rather than sporadic. I haven't gotten to addressing the Spanish and Portuguese outcomes of labials in this article yet, but Wireback 2002 ("On the Metathesis of Labials + /j/ in Hispano-Romance") thinks Spanish sepa underwent irregular metathesis, with analogical influence from other kinds of verb stem alternations. So while not a borrowing, an explanation for its development has been proposed.--Urszag (talk) 19:42, 21 September 2023 (UTC)
 * If a special explanation is needed I might suggest—in light of the fact that all three of the French words in question originally ended in /eV(m)/— that since hiatic Latin /e/ turned to /j/ later than /i/ did (the second change is attested earlier, in Classical poetry) their effects on preceding consonants could slightly differ. But, you know, original research and all that. Nicodene (talk) 20:26, 21 September 2023 (UTC)

/ks/ and /kt/
I'm thinking of merging these sections (and expanding the information on Albanian and Romanian outcomes). The outcomes of coda /k/ in these sequences are fairly similar within each branch of Romance, and for that reason they are often grouped together in when discussing sound-changes.

Thoughts, @Urszag? Nicodene (talk) 18:49, 27 September 2023 (UTC)
 * The next edit I am planning to make is an expansion of the start of the "other consonant sequences" section to add an overview of the dispute between the model Zampaulo advocates (where a change of coda velar > /j/ was the initial step in the palatalization of all of /gn kt ks kl gl/ in Western Romance) and the models described by Recasens or Baglioni where the apparent parallelism of these outcomes in Western Romance is considered to be deceptive in at least some cases. Because of that controversy, I wanted separate sections at least for /gn/ and for /kl, gl/ (which should also I think be expanded to cover labial + /l/ and consonant + l sequences in word-initial position) compared to /kt/ and /ks/, and I guess at that point I viewed it as simplest to have a separate section for each category of consonant that could follow the velar. But I think the repetition/separation of the portions of these sections describing non-palatalized outcomes in the current article is not ideal. I think the best options to fix that are either combining them into one section, as you propose (Repetti covers all of -gn-, -kt-, -ks- in one section), or revising them so that they describe just the palatalized outcomes, and moving the discussion of the Romanian outcomes of /gn/, /kt/ and /ks/ (which I think it is not controversial to view as parallel developments) to the introductory section where they can be treated together. I also plan to add to this initial section a table based on Baglioni pages 6 and 11 that shows the outcomes of ct, x, gn in different languages.


 * I think /kt/ and /ks/ mostly show parallel outcomes (with the possible exception of Tuscan, where several sources seem to consider it possible that [ʃʃ] could be a native outcome of /ks/ in some cases), so I have no major reservations about treating them together in one section. The one hesitation I might have is that I'm not sure whether Recasens views it as more likely that they developed in a parallel manner, or by somewhat different paths (since he says "A greater reluctance for consonants to overlap articulatorily in stop+lingual fricative sequences than in sequences composed of two stops (Byrd, 1996) ought to have implications for the interpretation of the causes of sound change", it seems he might consider it possible that /ks/ palatalized by vocalization while /kt/ underwent blending).--Urszag (talk) 23:35, 27 September 2023 (UTC)
 * I see. All things considered it may be simplest to have two new primary sections, one for intervocalic velar + /C/ sequences (/-ks kt kl ɡl ɡn-/) and the other for all other /Cl/ sequences, including initial /kl ɡl/. The latter section could include a note saying 'for intervocalic /-kl ɡl-/ see [link to previous section]'. Nicodene (talk) 22:31, 28 September 2023 (UTC)
 * That may be a good idea. There doesn't seem to be a perfect place to put discussion of palatalized intervocalic /kl/ /gl/ since there seems to be a fundamental disagreement about whether its original stages were connected to or distinct from the development of palatalized outcomes in postconsonantal and word-initial /kl/ /gl/ clusters and in other C + /l/ clusters.--Urszag (talk) 01:58, 29 September 2023 (UTC)
 * I know I've been slow on updating this section; I still have more reading to do before I have a sufficient understanding of the details and arguments on both sides. Currently I still lean towards thinking your original suggestion of merging the /ks/ and /kt/ sections might be best (but keeping /gn/ and /gl/, /kl/ separate). More information about Albanian and Romanian would certainly be welcome. I think -s- is also an attested outcome of -ps- in Romanian, isn't it (as in scrīpserāmus > scriserăm), and so -ks- > -ps- (> -ss?) -s- would be possible as an alternative to direct -ks- (> -ss?) > -s-; that might be worth mentioning.--Urszag (talk) 17:51, 25 October 2023 (UTC)

DIRECTU(M), etc.
Do we need this? The bracket-spam is a bit of an eyesore, and the loss of final /m/ is irrelevant to palatalization. Nicodene (talk) 09:09, 25 October 2023 (UTC)
 * I'm not totally sure, here are my thoughts:
 * I feel like using a standard format throughout the article is sufficiently important for readability to justify converting the cited forms into a single standardized format.
 * But of course, misrepresenting sources is not acceptable. This has been on my mind, but even so, I have not been sufficiently rigorous about indicating when alterations have been made. I'm not sure what the best way to handle this is in combination with the above.
 * Sticking "(M)" at the end felt at the time like a reasonable compromise between the different notations used by the sources: e.g. Boyd-Bowman (SŎMNĬC(Ŭ)LU), Zampaulo (CALCANEU) vs. Burger (lāneum), Repetti (LINEUM, UINEAM). Your edits to the -A-final words where I mechanically made that conversion have highlighted that this isn't as straightforward in the case of Romance feminine singulars that developed from Latin forms that were originally neuter plural. I initially rejected the option of CALCANEU(M), BURRIONE(M) vs. VINEA (which I think I've seen used in some papers) because it seemed inconsistent, but now I feel it would be an improvement over CALCANEU(M), BURRIONE(M), VINEA(M) vs. FOLIA/?FOLIA(M).
 * Formatting all Latin etyma like CALCANEU, BURRIONE, and UINEA is certainly a reasonable option. The main concern that I have about using forms like CALCANEU is that it seems it will be unobvious to readers without a background in Romance linguistics what form it is, and Latin words generally don't appear in this form in standard resources about Latin, such as dictionaries. Forms like BURRIONE seem potentially misleading in that they look like the Latin ablative singular form: although it can be argued the Latin ablative singular forms of words like this do partly underly Romance outcomes, it seems worthwhile to make sure that the reader is aware that the Latin accusative-case form would lead to the same outcome, and is commonly viewed as the source of the Romance form. All this could be fully explained in a footnote, but I worry some readers might miss it. By the same logic, I am uncomfortable about using just one footnote in an early portion of the article to indicate how forms attributed to cited sources have been altered. It seems like it would be ideal to have an unobtrusive link to this information at each transcription, but I'm unsure if there's a good technical solution for that (one footnote with multiple note-links would surely get unwieldy).
 * Another problem in this area is the representation of long vowels. Since this article isn't about vowel evolution, and the sources do not consistently mark vowel length, I've tried to generally standardize in favor of the omission of macrons in cited Latin forms, although I'm not sure if it would be more convenient in some places for it to be present.--Urszag (talk) 18:23, 25 October 2023 (UTC)
 * When I wrote the above I actually had in mind simply removing the brackets, which gives us CALCANEUM and so on. Standardizing the spellings found in our sources to the normal Latin ones seemed like a safe option to me, as well as the only one that wouldn't need to be explained in a footnote. I would prefer this to CALCANEU/BURRIONE/UINEA for the same reasons that you have elaborated.
 * As for the macrons, I'd actually thought of removing them during the latest 'copy-editing' but then thought I should probably ask you first before doing so. I agree that the Latin vowel length is largely irrelevant for our purposes here and doesn't need to be mentioned except on occasions such as the discussion about Italian legno vs. gramigna. Nicodene (talk) 18:49, 25 October 2023 (UTC)
 * Ah, I see! That also seems like a reasonable form to standardize on. I do think that it should be noted somewhere on this page that we have normalized the forms of Latin etyma to the standard Latin spellings, since otherwise--even if the difference is of no consequence--we'd inaccurately appear to be attributing the M-spellings to our cited sources. (I realize that the current version of the page already has the same problem with its use of (M) spellings.)--Urszag (talk) 19:11, 25 October 2023 (UTC)
 * Fair enough. I'll make the edit shortly. Nicodene (talk) 19:19, 25 October 2023 (UTC)
 * Shall I also go ahead and remove macrons where they are not relevant to the sound-changes being described?
 * Incidentally I agree that an easily-missed footnote may not be the best place for such information but don't see an easy alternative. I might've opted for a custom banner at the top of the page but I don't think that is normal practice.
 * Nicodene (talk) 19:39, 25 October 2023 (UTC)
 * Yes, sounds good to go ahead with removing macrons! You can do it or I will do it on my next article edit.--Urszag (talk) 17:58, 26 October 2023 (UTC)

'Conditions leading to palatalization'
Paragraphs 1–3 seem redundant as the information in them is conveyed elsewhere already. Perhaps they can simply be removed.

Paragraph 4 is interesting and, once properly cited (probably somewhere in Pope 1934), could serve as part of a new section that discusses the scholarly disagreement over phonemic palatalization in Proto-Romance, as discussed in Operstein 2010 Consononant structure and prevocalization (p. 107). Alternatively that can be addressed on our main page for Proto-Romance, since that is a matter of synchronic phonology whereas the focus here is diachronic.

Nicodene (talk) 14:39, 25 October 2023 (UTC)
 * Now that the article has been expanded a bit, and the original table has been removed, the section is somewhat of a relic. In terms of article architecture, my thought was that it is beneficial to overview some general information before going right into the coverage of how specific segments developed. E.g. while paragraph 2 might be belaboring the point, I think it is useful to tell the reader early on that palatalization in Romance languages did not occur in one stage, and to explicitly state this rather than just leaving it as something that can be inferred from the descriptions of the individual sound changes. Maybe a shortened version of this information could be put in the lead. As for paragraph 3, since the change from plosive > affricate > fricative is encountered in a number of different contexts, it also seems like it might be useful to familiarize readers with it so that they can interpret outcomes like TJ, CJ, CE, I > s in the context of there likely being an intermediate affricate stage. But I'm more indifferent to removing that paragraph; or alteratively, it might be better to make that point and give those examples at the end of the preceding section about the meaning of "palatalization".


 * Paragraph 4 is from the original article; I mostly left it as is because I haven't had a chance to follow up on it yet. Actually, one of the original motivations I had for working on this article was to follow up further on questions I had about the reconstructibility of palatalization in Proto-Romance, so I'd definitely be interested in seeing that expanded on, but perhaps as you said it will fit better and have more space to be done justice at the Proto-Romance article.


 * In its current form, this section definitely doesn't read perfectly, with the first paragraph being awkwardly duplicated by the next section below (although I think the "Development of Latin postconsonantal /j/" section should eventually be expanded to have more specifics). If you would like to rework it please feel free!--Urszag (talk) 17:51, 25 October 2023 (UTC)
 * We could drive the point home in the lead with something like '...a number of different sound changes that occurred, at various times and in various places, either in Late Latin [footnote here about Proto-Romance/Vulgar Latin] or its descendants...' And the end of section 2 does seem like a good place to mention the common plosive > affricate > fricative pathway.
 * I'll see whether I can unearth any sort of detailed debate about the phonemic status of palatalization. If not, then we might not be able to write much more than 'there is a disagreement' without engaging in Original Research. Nicodene (talk) 19:16, 25 October 2023 (UTC)
 * Update: I have not been able to locate such a debate, at least not in the Anglophone literature on this topic. For some relevant sources, the first footnote here can potentially be copied over. Nicodene (talk) 17:24, 9 November 2023 (UTC)

"Before front vowels" wording
Thank you Nicodene for the edits to streamline the article! There's one bit of wording I want to push back on: I don't like having the pre-I, E context generalized as "before front vowels" without any further specification of the vowels involved. I get that Repetti's use of /ki, ke, kɛ/ and /gi, ge, gɛ/ can be nitpicked since not all Romance languages show that specific three-way contrast between non-low front vowels, but I don't think the inaccuracy in this context is a serious issue, and I think it is useful to be clear that /ka, ga/ are not included because, as noted in the section on those sequences, some accounts of the palatalization of /ka, ga/ (such as Repetti's) view it a result of /a/ becoming a low front vowel in the languages involved (which is why Repetti includes it in section 39.3 "Consonant + front vowel palatalization"). Even though that explanation of the /ka, ga/ palatalization is seemingly debatable, I think it's a common enough viewpoint that "before front vowels" seems undesirably ambiguous to me.

I'm not wedded to Repetti "/ki, ke, kɛ/ and /gi, ge, gɛ/", although it's the best option that I can think of right now + easy to cite. If another source uses wording like "before high or mid front vowels" or "before Latin ⟨e i ae oe⟩" I guess something like that could be an alternative way of putting it.--Urszag (talk) 17:58, 26 October 2023 (UTC)


 * Very true. I don't subscribe to the '/a/ was a front vowel' theory myself so this hadn't even occurred to me. Nicodene (talk) 18:25, 26 October 2023 (UTC)

Suggestions
Hi, @FULBERT. Any suggestions for what could be improved? Other than the need for citations in the orthography section. Nicodene (talk) 16:24, 20 January 2024 (UTC)


 * @Nicodene Only some more References/Citations and perhaps some more incoming/outgoing links to other Wikipedia articles. Otherwise, I think this article is overall in nice shape! Thanks for all your contributions to it. FULBERT (talk) 16:31, 20 January 2024 (UTC)
 * Noted, thank you. Nicodene (talk) 16:32, 20 January 2024 (UTC)

/v/
Hi @Urszag. For the labial section we imply a Latin /v/ for what was rather /β/. Perhaps we could change the representation to /β/, with an explanatory footnote (+ source) noting its development from the Classical /w/ and /-b-/? Nicodene (talk) 19:02, 21 January 2024 (UTC)
 * I don't think /v/ is definitely wrong, since it is a phonemic not phonetic transcription and so /v/ doesn't necessarily need to represent a labiodental fricative phone (as opposed to "the sound represented by the letter V"), but I agree it may be misleading. Boyd-Bowman refers to "BY and VY (both [βj] in V. L." and Repetti to "/pj, bj, vj/".--Urszag (talk) 23:43, 21 January 2024 (UTC)

Latin /ɡ-/ + front vowel in Picard
The outcome is /ʒ/ and not an unpalatalised (!) /ɡ/, which would have placed Picard among the likes of Dalmatian or Sardinian. The /ʒ/ can be seen on ALF maps such as 638 or 639 (searchable here).

I see that the cited table does indeed show /ɡ/, but this can only be a misprint. What is the process for correcting this? Should we email Recasens and add his response here in a footnote? Nicodene (talk) 19:44, 21 January 2024 (UTC)
 * I also noticed this but have been putting off thinking about how to more generally verify and find examples for the results in the tables, while still keeping them an accurate representation of our cited sources. Unpublished personal correspondence is not citable as a reference on Wikipedia. If there was some kind of published erratum, we could cite it, but that seems a stretch to hope for (although "published" doesn't necessarily mean traditionally published, just "made available to the public in some form" by the source). But any response from Recasens would at least help to clarify that it was a misprint.--Urszag (talk) 23:28, 21 January 2024 (UTC)
 * I have tentatively added a footnote with more modest phrasing than I used here.
 * Incidentally I think I have been a bit cavalier with adding uncited (though correct) information to the orthography section, even if it replaced equally-uncited and less-complete information. I will work on fixing this. Nicodene (talk) 14:06, 22 January 2024 (UTC)
 * Also perhaps a bit cavalier with condensing or streamlining the article. (I mean no disrespect in doing so - the removed words include ones that I'd written myself.) Please feel free to revert any removals that you judge counterproductive. Nicodene (talk) 14:12, 22 January 2024 (UTC)
 * No problem, it wasn't very well cited beforehand and I was definitely thinking that the table taken from the Romance languages article needed some work. The changes look good to me so far but I may do more rereading and see if I have any suggestions. This is not important at all, but I slightly prefer over  since I find the latter to be on the uncomfortably small side. I haven't tried fiddling around with it yet, but given the function of the formatting I wonder whether it would be preferable technically (if possible) to use {{lang|la|... and have the template apply some style that would format the text appropriately. But the section on styles at Template:Lang seems to only talk about personal settings, not how to set a default style for all users within one article.--Urszag (talk) 04:52, 23 January 2024 (UTC)