As my co-presenter Jorik Groen and I both sadly couldn't attend #NACAL47, @PhDniX just did us the great favour of presenting on our behalf. Here's a thread with our slides. For the references and to give us your (non-tweeted) feedback, please visit http://tinyurl.com/nacal47 !

Our talk is about #Semitic nouns with 3 consonants and a vowel in the singular stem, like *kalb- 'dog'. (C stands for any consonant, V for any vowel). In some langs, their plural inserts an *a-vowel & also takes pl. endings. This is seen as an innovation, but we think it's v old.

This is the outline of our talk. 1st, we argue that this double marking (a-infix, plural suffix) of *CVCC- plurals could be Proto-Semitic. 2nd, that an ancestor of PS had a plural suffix *-w- that gave rise to the PS plural suffixes. 3rd, that this explains the a-infixation.

Here is a more-or-less conventional family tree of #Semitic, based on https://www.academia.edu/2603558/Phyla_and_Waves_Models_of_Classification_of_the_Semitic_Languages. Languages indicated with red attest the doubly-marked plural forms we're talking about, as we will see.

In the Northwest Semitic languages, double marking is the rule. Although these examples all have an *a vowel in the stem, *CiCC- and *CuCC- nouns also insert an *-a- in the plural. Same goes for words with the *-at- feminine suffix.

But we also find traces outside of Northwest Semitic. In #Arabic, this is a common pluralization pattern for CVCC-at- nouns, while we find an isolated form in the word for 'country, land'; also in #Geez 'dog'.

The #Arabic and #Geez forms don't look like recent innovations (or borrowings). In other branches of #Semitic, we either lack information about the (historical) vocalization of these plurals, or sound changes would have covered the evidence up. So this pattern could well be PS.

So what about those plural endings? This is how we can reconstruct them for Proto- #Semitic; the forms are continued unchanged in (Babylonian) #Akkadian, #Ugaritic, and Classical #Arabic, for instance. Can we reconstruct even older, pre-Proto-Semitic forms? (spoilers: yes)

Dense slide. The pattern of long and short vowels in the plural endings is reminiscent of a few kinship nouns. Here, it has been explained as deriving from a sound law that deleted *w and *y between a consonant and a vowel, lengthening the vowel. Could also work for the pl endgs.

Evidence for *-W- as a plural ending also comes from non- #Semitic #Afroasiatic, especially Ancient #Egyptian. @PhDniX is angry with us for leaving out #Berber; there is also definitely relevant evidence there, but it's complicated, so we hushed it up for now.

The Egyptian plural endings match the internal pre-Proto-Semitic reconstruction. Especially the weird order of number-gender suffixes (plural-feminine) in both must be related (normally, languages mark gender closer to the stem than number). So: pre-PS pl. suffix *-w-.

So, we've reconstructed back beyond Proto- #Semitic, but we still have these redundantly marked plurals: if the number is already marked by a suffixed *-w-, why also infix the *-a-? And why in *CVCC- nouns in particular?

What's different about the *CVCC- nouns? They end in two consonants. Add the plural *-w- and it becomes hard to pronounce without the *-a-. So we propose that the *-a- is an epenthetic vowel, inserted to make the sequence of three consonants easier to pronounce.

We can check this by looking at other nouns which some people (including me) reconstruct as consisting of 2 consonants and 0 vowels, such as *bn- 'son' (see my handle). Our epenthesis rule predicts that we should get *-a-insertion + pl endings there too. And for *bn-, this holds!

In conclusion, we suggest that the double plural marking in these nouns is the result of sound change and not of something like the broken plural system of langs like #Arabic, which change a noun's stem to form the plural. Hence, we have nearly no evidence for broken pls in NWS.

That's it! Thank you for reading; I hope Twitter is just as excited about reconstructing pre-pre-Proto- #Semitic as we are. Again, we'd love to hear your questions and comments, either here or at http://tinyurl.com/nacal47 .