Oscan Reconstruction

[marconatrix]

Re. *seks* would /ks/ be a valid cluster, or would it have developed something like /kt, pt/ as happened in Celtic?
Also from Celtic, I haven't time to dig into this now, but your forms of 'four' with and without a /w/ remind me of the Welsh (and Cornish and Breton) m. and f. versions of this numeral, _pedwar_ (m), _pedair_ (f).
Love the idea of a recon p-Italic btw.

[kanejam]

kanejam, if you haven't already, I recommend looking at Andrew Sihler's "A New Comparative Grammar of Greek and Latin". It's an updated version of Carl Buck's and while it's focused on the histories of Greek and Latin, he does draw in Oscan material and comment on Proto-Italic forms.

I actually have that on my computer but had forgotten about it and hadn't read it yet. Thanks for reminding me about!

I couldn't resist commenting that the noun paradigms keep reminding me of Latvian, which a looked into a bit a while back. Sort of the same level of deviation from PIE, simplification of endings and so on. Obviously a parallel development but maybe of interest to someone?

I've never looked into Latvian so I can't comment at all.

Re. *seks* would /ks/ be a valid cluster, or would it have developed something like /kt, pt/ as happened in Celtic?
Also from Celtic, I haven't time to dig into this now, but your forms of 'four' with and without a /w/ remind me of the Welsh (and Cornish and Breton) m. and f. versions of this numeral, _pedwar_ (m), _pedair_ (f).
Love the idea of a recon p-Italic btw.

You're right that *seks should follow the rule of /ks/ -> /ss/, giving us sess instead. I can't remember now why I thought that pettiur should inflect; maybe to explain the petiro- form and also I think I read Buck mention something about *k^wettwōr still inflecting in Proto-Italic.

[kanejam]

Whoops, double post...

[kanejam]

I haven't been working too hard on this recently, sorry to whoever was following this. But I promise this is still on my mind and I'll get back to it soon. But as a bonus I just discovered wiktionary's reconstruction of the Proto-Italic noun system, and while it's incomplete and I don't agree with everything they've written it has given me an idea. I will have a quick go at the 4th declension, but not just what they would look like if they descended straight from Pr-It but what they are more likely to be given the scant forms we have and the blurring from the i-stems that is suggested.

4th Declension

SG
Nom ... -s ... mans ... -s ... *-us
Voc ... -s ... mans ... -s ... *-us
Gen ... -úvs ... manúvs ... -ūs ... *-ous
Dat ... -veí ... manveí ... -uī ... *-owei
Acc ... -im (-um) ... manim ... -um ... *-um
Abl ... -id (-ud) ... manid ... -ū ... *-ūd
Loc ... -ou ... manúv ... -ū ... *-ou
PL
Nom/Voc ... -s ... mans ... -ūs ... *-?
Gen ... -um ... manum ... -ūm ... *-wom
Dat/Abl ... -uss ... manuss ... -ibus ... *-ufos
Acc ... -uss ... manuss .. -ūs ... *-uns

The dative singular might drop that -v- with the pressure from the i-stems and the plural oblique forms might well be -íss rather than -uss. The ablative singular would probably end up as -eí after the i-stems as well. If all of the above happens, and the assumption of the nominative plural is correct, we end up with basically an i-stems which has an irregular genitive singular.

I'm still going to ignore the 5th declension, and assume if I have to that they too will have merged with the i-stems.

Edit: Unfortunately wiktionary has no info on verbal declensions and de Vaans' etymologies don't worry too much about inflection either. Hopefully sometime I'll finally get round to having a look through the comparative grammar of Greek and Latin.

[Tropylium]

A question: de Vaan in Etymological Dictionary of Latin and the other Italic languages presents a very brief outline of the historical phonology of Latin. However, he does not comment at all on the other Italic languages. Given his list of innovations, which of them can we clearly confirm per Oscan evidence (or Umbrian etc.) as Proto-Italic?

3.1 *tˢt > *ss
3.2 *CLHV > CaLV- (e.g. calēre, valēre, palma)
3.3 *CNHV > [CəNV-] (e.g. similis, sine, tenuis)
3.4 *HLC- > aLC- (argentum)
3.5 *h₁ ₂ ₃NC- > e/a/oNC- (umbicīlius, ambi)
3.6 *p_(R)kʷ- > *kʷ_(R)kʷ- (coquō, quinque)

4.1 *bʰ, *dʰ, *gʰ, *gʷʰ > *b, *d, *g, *gʷ after *s and *N
4.2 *gʰ > *g before *l, *r (Lat. glaber, grāmen, tragula)
4.3 *bʰ, *dʰ, *gʰ, *gʷʰ > *f, *þ, *χ, *χʷ
4.4 PIE *ouV > *awV in pretonic position (Thurneysen-Havet-Vine's law) (Lat. lavō etc.)

5.1 *þ- > *f- (Fal. filea, Lat. filius)

6.1 *fw- > *f- (Lat. forum, fores)
6.2 *-gʷ-, -χʷ- > *-w- / V_V (Lat. voveo, nūdus)
6.3 *mj > *nj

7.1 *o > a / b, l, m, w, kʷ_CV (Lat. badius, canem, lacus, lanius, manus, mare)

8.1 *ew > *ow (Lat. novus, moveo)
8.2 *r̥, *l̥ > *or, *ol (Lat. morior etc.)

He comments that Venetic shows innovations 4.3, 5.1, 8.1 and 8.2, and takes this as an indication that Venetic only branched away at stage 8, therefore implying that Sabellic brances away later yet. But I don't actually see how each stage here would both group together, and be distinct from each other; e.g. stages 5 and 6 appear to be entirely unrelated to stage 7.

[kanejam]

It's been a while since I've done any work on this and I'm pretty busy at the moment so a decent reply won't be forthcoming any time soon.

I can confirm the changes 3.1, 3.4, 4.1, 4.2, 4.3 (and I think 4.4 as well), 5.1 and 8.1. I know that gʷ -> b as part of the large labiovelar shift (and I'm pretty sure χʷ -> f as well). I'm unsure about the others, but it certainly seems that they share plenty of changes.

About 4.3: I've been entertaining the notion that the aspiratae actually became voiced fricatives that devoiced initially, rather than voiceless fricatives that voiced intervocalically.

[Sumelic]

It's been a while since I've done any work on this and I'm pretty busy at the moment so a decent reply won't be forthcoming any time soon.

I can confirm the changes 3.1, 3.4, 4.1, 4.2, 4.3 (and I think 4.4 as well), 5.1 and 8.1. I know that gʷ -> b as part of the large labiovelar shift (and I'm pretty sure χʷ -> f as well). I'm unsure about the others, but it certainly seems that they share plenty of changes.

About 4.3: I've been entertaining the notion that the aspiratae actually became voiced fricatives that devoiced initially, rather than voiceless fricatives that voiced intervocalically.

Are there any particular reasons to think this? It seems less likely to me. For one thing, in European languages in general, including the languages descended from Latin, intervocalic voicing of voiceless fricatives seems to be a more common process than devoicing of word-initial fricatives. For another thing, there is at least one example that seems to clearly show intervocalic voicing of voiceless fricatives in the history of Latin: the rhotacism of intervocalic "s" (Wikipedia says written evidence of this occurred in the transition from Old Latin to Classical Latin, around 350 BC, but /r/ is assumed to be derived from an earlier Old Latin [z] allophone of /s/).

In terms of phoneme inventory, something like /p b f t d θ k g x/ seems a lot more likely to me than /p b v t d ð k g ɣ/. It seems to me that for most languages with non-sibilant fricatives, a certain POA either has voiced-voiceless pairs for both plosives and fricatives (like Irish or modern Greek /k g x ɣ/), or has voiced-voiceless pairs for plosives but only a voiceless fricative (Standard German and Russian, for the velar sequence: /k g x/), or has phonetic voiced fricatives that don't contrast phonemically either with voiced plosives (Spanish, and from what I can tell Gothic /k g~ɣ x/) or with voiceless fricatives (apparently Old English, for the coronal sequence: /t d θ~ð/) or both (it seems to me this is the case in languages like Finnish or Inuit languages that have things like /p v/, /k ɣ/ or /q ʁ/). For example, looking at the Wikipedia page for the voiced velar fricative, I notice that there are a ton of notes saying it is an allophone of /g/, or in some cases, /x/.

[Sumelic]

I thought of a sort-of exception to what I said above: there are a fair number of languages I can think of with voicing contrast in labial plosives, no native /f/, but that have something like /v~ʋ~w/. I don't think any of them count as a clear-cut /p b v/ though; the general pattern is that/v/ is derived from and still patterns to some extent as a semivowel, and /f/ is well-established from loanwords or other sources. Certainly none of them have a four-way contrast between /p b v w/ without /f/, which would be necessary to get the four distinct Classical Latin reflexes /p b f w/.

Hindustani/Hindi/Urdu: has /p pʰ b bʱ ʋ/, and also /f/, but /f/ only occurs in borrowed words, where it has historically sometimes been realized as /pʰ/ (Wikipedia suggests that most modern speakers are capable of articulating [f] though).

Many Slavic languages have /p b v f/ but with /f/ only being found in loanwords (although these words are usually numerous and well-established). The phoneme /v/ usually derives from earlier [w] and may still be realized this way in some contexts in some languages (such as Ukrainian and Belarusian).

Modern Hebrew has /p b f v/ but /f/ in native words doesn't contrast with /p/ at the start or end of a word. However, there are loanwords with word-initial /f/ or word-final /p/, and intervocalically the two are contrastive in inherited words due to the loss of consonant length.

So actually, all of these languages still have /f/ in some way. Also, none of them resolved the instability of /p b v/ by devoicing word-initial /v/.

The only example of word-initial devoicing of fricatives I can think of is in German (for /z/) and Dutch (for /z/ and /v/), where it appears to be a reversal of earlier general voicing of word-initial fricatives /s/ and /f/. I'm not even sure if this is a true devoicing shift, or if it's a case of the relevant dialects never applying the voicing shift and just retaining original [s] and [f].

[Travis B.]

The only example of word-initial devoicing of fricatives I can think of is in German (for /z/) and Dutch (for /z/ and /v/), where it appears to be a reversal of earlier general voicing of word-initial fricatives /s/ and /f/. I'm not even sure if this is a true devoicing shift, or if it's a case of the relevant dialects never applying the voicing shift and just retaining original [s] and [f].

At least in the case of German, initial [z] is the innovation, probably due to the influence of Low German, and initial [s] is the conservative form.

[kanejam]

I don't have any great big reason to think this. My thinking is mostly in two parts: firstly the question of how the voiceless fricatives get there in the first place. AFAIK all theories of the phonology of PIE place the aspiratae as voiced stops, so a simple lenition to voiced fricatives seems more likely. Secondly, I think the idea of voiceless fricatives is a carry over from the days when Greek and Latin were thought to be closely related and the fricatives were thought to have come from voiceless aspirates, but now there is no reason to posit voiceless aspirates at an intermediate stage, especially if you don't think the aspiratae were breathy voiced (which I don't - at least not until Greco-Armeno-Aryan).

You're right to point out that the resultant series /p b v w/ is really unlikely, and that intervocalic voicing is already attested by intervocalic /s/ -> [z] (which happened already in Proto-Italic). My thinking is that the initial fricatives devoiced straight away due to the unstable nature of the series, and that afterwards the fricative series was phonemically voiceless (or at least unmarked for voicing) without ever having been realised as such intervocalically. It's not a particularly convincing argument I know.

[Sumelic]

I don't have any great big reason to think this. My thinking is mostly in two parts: firstly the question of how the voiceless fricatives get there in the first place. AFAIK all theories of the phonology of PIE place the aspiratae as voiced stops, so a simple lenition to voiced fricatives seems more likely. Secondly, I think the idea of voiceless fricatives is a carry over from the days when Greek and Latin were thought to be closely related and the fricatives were thought to have come from voiceless aspirates, but now there is no reason to posit voiceless aspirates at an intermediate stage, especially if you don't think the aspiratae were breathy voiced (which I don't - at least not until Greco-Armeno-Aryan).

You're right to point out that the resultant series /p b v w/ is really unlikely, and that intervocalic voicing is already attested by intervocalic /s/ -> [z] (which happened already in Proto-Italic). My thinking is that the initial fricatives devoiced straight away due to the unstable nature of the series, and that afterwards the fricative series was phonemically voiceless (or at least unmarked for voicing) without ever having been realised as such intervocalically. It's not a particularly convincing argument I know.

Hmm... Well, it may be slightly similar to Old English, which is analysed with /p b f~v w/. The sound [v], considered an allophone of /f/, descends in some cases from Proto-Germanic *f but in other cases from the *ƀ allophone of Proto-Germanic /b/, so it might be considered in these cases to have changed phonemically from /b/ to /f/ without ever actually losing voicing. Word-initially, of course, /b/ and /f/ remained distinct.

It also sort of reminds me of the Modern Greek spirantization of /b d g/, which did affect word-initial consonants.

So I guess it comes down to which seems more probable: general devoicing of PIE aspiratae followed by general spirantization followed by intervocalic voicing, or general spirantization followed by word-initial devoicing. The first sequence has more steps, but each step seems fairly probable to me. (Unconditional devoicing of a series of voiced stops, even if tenuis rather than breathy-voiced, seems OK to me; I think there are examples from languages like Chinese or Tibetan.) For the second sequence, the sticking point for me is the conditional word-initial devoicing.

[Richard W]

I think there is a perfectly reasonable explanation of the development of the 'voiced aspirates' in Chapter 5 of Jane Stuart-Smith's Phonetics and Phonology: Sound Change in Italic. She starts from the shockingly conservative hypothesis that the phonations were as in Hindi, i.e. *t = [t], *d = [d] and *dʰ = [d̤], and looks at what happens to these sounds in Indic languages. She ends up with d̤ > tʰ initially (as in Romany - bhrātā > English _pal_ is the mnemonic example); also generally reconstructed for Thai and Lao and their minor neighbours and I think actually visible in the ensemble of Khmu dialects) and d̤ > ð between vowels. I don't remember her analysis of intervocalic clusters - their development across Italic is not trivial. She also motivates the anomalous development of initial *gʰr > gr- in Latin.

[Tropylium]

I thought of a sort-of exception to what I said above: there are a fair number of languages I can think of with voicing contrast in labial plosives, no native /f/, but that have something like /v~ʋ~w/. I don't think any of them count as a clear-cut /p b v/ though; the general pattern is that/v/ is derived from and still patterns to some extent as a semivowel, and /f/ is well-established from loanwords or other sources. Certainly none of them have a four-way contrast between /p b v w/ without /f/, which would be necessary to get the four distinct Classical Latin reflexes /p b f w/.

(…)

The only example of word-initial devoicing of fricatives I can think of is in German (for /z/) and Dutch (for /z/ and /v/), where it appears to be a reversal of earlier general voicing of word-initial fricatives /s/ and /f/. I'm not even sure if this is a true devoicing shift, or if it's a case of the relevant dialects never applying the voicing shift and just retaining original [s] and [f].

Old Irish might be another example, shifting initial PIE *w to /f/. But there doesn't seem to be hard evidence on if this was *w > *v > f or perhaps *w > ʍ > f. Brythonic does something else entirely by shifting *w- > *gw-, so that doesn't help.

[Richard W]

Another devoicing of (initial) fricatives that probably doesn't help is the Tai abandonment of voicing contrasts, where we see a merger of the plausible (initial) voiced series /v z ɣ/ with /f s x/. (Preɡlottalised stops then generally simplified to voiced.)

[Lambuzhao]

I enjoy this thread and the reconstructions you came up with. I can't tell you how many times I pored over Buck concerning Oscan and Umbrian and just slobbered at the "alien yet familiar", almost Middle Earth-like qualities of these two langs (Tolkien missed a doozy of an opportunity with these bears!). I am itching to find some old notes where I did some reconstructions, and 'compare notes', as it were. But on the whole, really interesting work.


I did have one little question, though. You might have covered this, b/c you covered so much, but here goes.

On the one hand

This is probably enough for one go, but here is a reconstructed present indicative of faamat:

1SG - faamahu (that h is unattested but it would make sense)

Yet, on the other hand

The present infinitive ... tríbarakavúm 'to build', with the v being merely a glide.

Don't get me wrong, *faamahu quacks like an Oscan, as far as I'm concerned. But, wouldn't *faamavu, or maybe *faamavú, be closer, since the /v/ glide is attested in that environment?

[dhok]

For the supposed instability of /p b w v/, note that modern Polish has this system except for /f/ from loanwords.

[hwhatting]

For the supposed instability of /p b w v/, note that modern Polish has this system except for /f/ from loanwords.

Not only. There is at least one case of /f/ from native sources (ufać "trust" from upwać), and in the dialects /f/ from /xw/ is frequent, resulting in native toponyms with /f/. IMO, that only goes to show that a system /p b w v/ is indeed unstable and likely to both develop /f/ on its own and to easily integrate it via loan words.