Reconstruction talk:Proto-Semitic/ṯalāṯ-

Page moved instead of being deleted

@Profes.I.: Next time you find a Reconstruction page with a wrong title, it's important to use the "move page" function to change the name rather than starting a new page and tagging the old page for deletion. This is important so that the page history can be preserved. Also, I've kept a redirect from the old name since a large number of pages continue to use the old name, and it's better for them to have a redirect than a red link. —Mahāgaja · talk 14:41, 13 July 2020 (UTC)

Incorrect Reconstruction

@Fay Freak, Victar, Fenakhay, Profes.I., 334a, Wikitiki89, Metaknowledge Arabic and Northwest Semitic show regressive assimilation from *śalāṯ- > ṯalāṯ-. This type of assimilation is common among numerals (cf. PIE *pénkʷe > Celtic, Italic *kʷénkʷe, Pre-Germanic *pémpe). The same thing occurs in Ugaritic ṯiṯṯ- < *sidṯ- (“six”). On the basis of the reflexes outside of Arabic and Northwest Semitic, the Proto-Semitic form must be reconstructed as *śalāṯ-.

Compare:

• Old Akkadian: 𒊓𒇷𒅖𒁴 (sa-li-iš-tim /⁠śaliṯtim⁠/) (genitive feminine ordinal)
• Sabaean: 𐩦𐩡𐩻 (ślṯ) (early Sabaic)
• Ge'ez: ሠላስ (śalās)
• Mehri: śīləṯ (day counting form)

Rhemmiel (talk) 08:29, 14 August 2020 (UTC)

Can we source these attestations for certain? I think it would be prudent of us to know if any or all of these are actually source-able before weighing them as evidence one way or another. You do bring up an interesting parallel of Ugaritic's handling of š-d-ṯ as ṯ-d-ṯ, but as can be seen there is great variety in how the phonemes atypically shift for that word in the Semitic family, as well as even for the reconstructed Proto-Afroasiatic root in its descendants. I do see Aaron Rubin reconstructing proto-Modern South Arabian as *ślṯ, this is where the mention of an Older Sabaic form is found glossed in a footnote of "Omani Mehri A New Grammar with Texts", quote "A feminine form ślṯ (vowels unknown) is attested in Early Sabaic and in the other OSA languages (though later Sabaic has ṯlṯ)." The later form being the more prevalent attestation in the available online corpora, but this might be deceptive as we have more texts from later periods than earlier. The source of this regressive assimilation theory is Lipinski in "Semitic Languages an Outline of Comparative Grammar" specifically note 35.8 under Numerals. My real trouble is with the Akkadian; I think it is an oversimplification of cuneiform orthography to associate a proto-Semitic sound scientifically with the use of one form of "s" over another "s". I've scoured everywhere for this variant form spelt with 𒊓 (sa) which is the major speculation point to his proposal; two attestation only, one by Sargon and one by his immediate successor Rimush. Many words are spelt with a great variety of forms far greater than this frequency and we think nothing of it besides scribal convention and discretion; that is the norm in standardized periods, how much more so reasonable in the earliest adapting of Akkadian language into cuneiform. What it looks like to me is only the South Semitic branch has *ś while the rest obey the expected pattern of *ṯ and *š. In reverse, perhaps it was they who did not tolerate the double appearance of *ṯ in a word and adopted an easier alternative for them. After all regarding assimilation, *ś is oftentimes reconstructed to have a lateral quality just as the middle radical /l/ in the word; South Arabian is the branch that preserves that pronunciation of the phoneme, while the other languages have lost it. My recommendation is to perhaps restructure the page to include the subcategory under West Semitic called South Semitic, remove the Akkadian variation entirely as it does not fairly depict the normalized form, and include Lipinski's theory by footnote. -Profes.I. (talk) 12:09, 14 August 2020 (UTC)

What Profes.I. said. He beat me to the punch. The spelling with 𒊓 (SA, ŠA) shows nothing at all like ś against š, so drop it. The Modern South Arabian languages – also Soqotri śə́lɛh, Shehri śhǝléṯ, śɔṯét as written by Johnstone – are attested only late and show even more distortion, and with the Sabaean one can equally well support *ṯalāṯ- because of it more often having been spelled 𐩻𐩡𐩻 (ṯlṯ). With thus small a corpus it may well be a coincidence that a dissimilation 𐩦𐩡𐩻 (ślṯ) appears early. It shows the Sabaeans had that dissimilation phenomenon but it did not catch on because of conservative scribes, whereas amongst the Modern South Arabian speakers nobody stopped them to make the lexeme more comfortable. The geographical area of these *śalāṯ--like forms is very narrow, just partially OSA and MSA and Ethio-Semitic which stems from Southwestern Arabia. Whereas to pose it as original you would need to assume that regressive assimilation took place in all the rest – Akkadian, Ugaritic, Aramaic, Canaanite, Arabic parallely? No way. Fay Freak (talk) 13:08, 14 August 2020 (UTC)

I add that cases where a Semitic language has an identical first and third root consonant are altogether rare. See Greenberg, Joseph Harold (1950) “The Patterning of Root Morphemes in Semitic”, in Word‎^[1], volume 6, number 2, →DOI, pages 162–181, Vernet i Pons, Eulàlia (2011 March 1) “Semitic Root Incompatibilities and Historical Linguistics”, in Journal of Semitic Studies, volume 56, number 1, →DOI, pages 1–18, Vernet i Pons, Eulàlia (2016) “Etymologischer Ursprung der reduplizierten und geminierten Wurzeln im Proto-Semitischen”, in Proceedings of the 6th Biennial Meeting of the International Association for Comparative Semitics and Other Studies (Babel und Bibel; 9)‎^[2], Winona Lake, Indiana: Eisenbrauns, →DOI, →ISBN, page 193. From this one can derive that there has been a tendency to eliminate these constellations, so we are reconstructing an archaism here, retained in certain common words as they are less liable to suffer arbitrary sound substitutions. Fay Freak (talk) 03:53, 28 July 2021 (UTC)