I'm not sure if you're asking about the phonetic realisation or the phonology of it, but the syllable /aw/ behaves precisely the same as any vowel-consonant sequence. If you add a vowel, for example, the /w/ resyllabifies: /kaw.tus/ > /ka.wē.re/, and vice versa /a.wi.kel.la/ > /aw.kel.la/. Further, /aw/ was the only original diphthong that had survived into proto-Romance the multiple waves of diphthong eliminations; where it remains today, it's primarily (if not exclusively) as two syllables (probably the whole of Italy) or as a vowel-consonant sequence. Latin "short/lax" vowels can only exist as the syllable nucleus, and they get invariably deleted as the first vowel in hiatus. The only truly diphthongal area I can think of is Galician-Portuguese. The last pointer is quite straighforward: the /w/ in /aw/ never develops in the same way as /ŭ/. Granted, that's begging the question.So, you assume Paulus was /pawlʊs/ rather than /paʊlʊs/. Could you explain why?
Indeed, e.g., nubilus > nuvolo > nugolo, uvula > ugola.Certain (Tuscan) dialects shifted the hiatus-filler to /g/ (this interchange is already visible in Late Latin, e.g. in σαγμα /saɣ.ma/ > sauma /saw.ma/) and is otherwise common in Italy and the islands; others retained it as /v/; yet others dropped the hiatus filler altogether, presumably because they tolerated hiatus more, and this was selected as standard. Hence we get the three forms Pagolo, Pavolo and Paolo.
I agree. And likely with Paullus too, as in Catalan we've got two well-established variants, Pau and Pol.The Latin form in -au- coexisted with Pōllus, originally dialectal but very common in the city of Rome by the Late Republic. It doesn't seem to have survived into Romance, likely due to eventual homophony with the outcome of pullus.
It would be in absence of any other explanation. Tuscan Italian doesn't preserve the PRmc /a͜w/, but regularly monophthongises it to /ɔ/, as do many other Romance varieties. Paolo is clearly a special development, for which I gave a straightforward and regular explanation. It's crystal-clear that it contains the regular diminutive suffix.Well, exactly:
Paulus > Paolo would be such an example.
Paullus is the earlier form that underwent the regular degemination to Paulus, that applied after AV and AE, but not after long vowels. It is precisely after this happened (1 c. BC) that one can confidently treat /aw/ as a vowel-consonant sequence - a necessary condition for the shift. The fact that it applied after /ae/ (caelum < *caed-lom) coincidentally tells us that it wasn't monophthongised yet at that point. If geminates can only occur intervocalically, then the V and E in question weren't vowels, but consonants. The geminate spellings continued to exist as graphic archaisms (also in caussa, cāssus), but the grammarians explicitly say that they cannot be pronounced as spelt. Notice in cāsus that consonants other than /l/ were simplified also after long vowels.That would be a counter example agaist the phonological analysis <au>=/aw/ rather than /aʊ/. Geminate consonants can only occur intervocalically in Latin.
*paurelos is unattested and is merely a reconstructed form. Its development to Paullus via unstressed vowel deletion was complete by about the 5 c. BC, which pretty much excludes any such surivors. Latin has very regular dissimilation of r-r and l-l (which continues even in modern Castilian), but r-l sequences themselves are desirable and stable. There's neither a reason for nor an example of /r > w/, least of all with a /w/ already preceding. In this case the regular development is metathesis: *pawros > parwos.
The same as I gave for Italian: /paw.lũ/ > reanalysis /pa.(w)u.lũ/ > regular syncope /paw.lũ/ > regular B~V merger /pablo/. I'm not sure about the syllabification in Old or New Spanish, but the outcome being the same as that of tabula and different from causa testifies that /bl/ was homosyllabic when /aw/ monophtongised in Castilian. Curiously, Venetian seems to have both Pagoło and Poło, but only toła; it's the reverse in French (Pol, table). The syllabification of muta-cum-liquida seems to have remained unstable from pre-literary Latin right down to this day.By the way, what's the standard explanation of the phonetic evolution of Pablo?
What development exactly are you referring to?Your idea with w faces one major problem: the lack of examples of such a development in Latin.
Paullulus>paululus rather suggest that paullus>paulus weren't perceived as diminutives anymore, so what would be the motivation of this special development paulus>*pavulus?
All geminate /ll/ were prototypically palatalised. VILLA regularly spells /vīl.la/ like POLLVS spells /pōl.lus/.
Yes, this is precisely parallel to the change I'm talking about. In fact it could be the exact same change: consonant-initial suffixes were no longer felt to be the basic forms, replaced by vowel-initial ones: puer+lus > puer+ulus :: paw+lus > paw+ulus.
The last two are simply and openly an impossible syllable structure in Latin, or even ancient Indo-European as a whole. /u/ after a vowel and before a consonant surfaces as [w]. /r/ between two consonants surfaces as [r̥]. /Vrl > Vll/, or conceivably /V̄l/, but in no case /Vl/.*pau̯relos>*pau̯rlos>*pau̥rlos>*pau̥los
These would be short-lived intermediate stages following the syncope. *Agros>*agr̥s>ager acquired r̥ long after the PIE syllabic sonorants had vocalized in Proto-Italic.The last two are simply and openly an impossible syllable structure in Latin, or even ancient Indo-European as a whole.
And for all we know there was a prolonged stage with secondary syllabic sonorants. But what you're suggesting is equivalent to *agros > ag̥r. It's not a question of how long this stage lasted - it's that such a syllabification is without precident at any stage from PIE to modern Romance, which are all basically CV languages that demand a syllable onset, and as such it's synchronically impossible. Even if it was possible, *pau̥rlos > *pau̥los is impossible but must > paullos; Even if that was possible, the only syllabification of /paulos/ that is possible in any kind of Latin is /paw.los/, and postulating /pa.u̥.los/ is the same as postulating /mo.n̥.te/. The only way to get /pa.(w)u.los/ is through a morphophonemic metanalysis, which is what I've proposed.These would be short-lived intermediate stages following the syncope. *Agros>*agr̥s>ager acquired r̥ long after the PIE syllabic sonorants had vocalized in Proto-Italic.
Here it's worth asking why create such a cluster in the first place, when you have an already-existing solution in the metathesis? Ideally it should serve as an event horizon preventing us from speculating about what would have happened to the sequence had it not metathesized.But your fictitious examples imply a disappearance of the syllable, which is the opposite to what I suggest. In my scenario *u̯r̥l becomes (u̯)ul. Try to pronounce *pawrlos yourself several times and try to observe what happens in each case. I am sure *pawərlos and *pa.u.rlos and *pawəlos will emerge as variants.
The o>u raising isn't even surprising next to examples like ampulla < ampora. It's the back vowel itself in all the on-stems.
The s is between the stop and the l, and as far as I can see that s is completely transparent to the outcome. A lone s lengthens the preceding vowel (vīlis < *wes-li), as does stop + l (pālus<*paksl-). I've given three examples of diphthong + stop + l, all of which were phonemically no different from au. I think you already have what you're looking for.Yes, aulla etc. are good examples, but with s, which deletes everything between it and l, compare ex vs. ēlūdō. So, we need au + stop + l.
The advantage of my explanation is that metanalysis doesn't require being observed in a language in order to be postulated - it's a universal linguistic process. Of course, in our case this exact metanalysis has been noted in the very same language by your own self. Most importantly, we're not talking about phonetics, and therefore what exactly precedes our metanalized diminutive suffix simply doesn't matter. It makes no difference in the least whether it was paw- or puer-, in this is why I think my explanation simply can't be shot down.P. P. S. A development au̯>au̯u, even by analogy as you suggest, has not been observed in Latin either. We're dealing with a unique case, whatever the explanation.
I only see caelum<*caedlom, which by the way has no accepted etymology.The s is between the stop and the l, and as far as I can see that s is completely transparent to the outcome. A lone s lengthens the preceding vowel (vīlis < *wes-li), as does stop + l (pālus<*paksl-). I've given three examples of diphthong + stop + l, all of which were phonemically no different from au. I think you already have what you're looking for.
These latter involve s, which deletes the preceding consonant together with itself even in Slavic (luna<*lou̯ksnā, lono<*logʰsno(m) ~ ložesno<*logʰesno(m) vs. agnьcь<agnikos). I am not convinced about caelum.The "sky" one has a couple very convincing ones, both involving kaid-. But I'm talking about the "chisel" one. The other ones a lūna and vīlla.
The s doesn't delete the preceding consonant. The whole cluster is assimilated to l after a diphthong, or adds a mora to the vowel otherwise (not sure if there are examples after a clearly long vowel). Are you expecting some other development without the s?These latter involve s, which deletes the preceding consonant together with itself even in Slavic (luna<*lou̯ksnā, lono<*logʰsno(m) ~ ložesno<*logʰesno(m) vs. agnьcь<agnikos). I am not convinced about caelum.
AVGVSTA (latin) – ? – Agòsta – Avòsta – AòstaIt seems to me that the stressed "o" of "Aosta" results from the stressed "u" of "Augusta", not from the diphthong "au".
The case of Paulus > Paolo is different from this, isn't it?