There are many Indo-European roots with alt. of H2/3, sometimes also with other variants with H1 ... more There are many Indo-European roots with alt. of H2/3, sometimes also with other variants with H1 or 0. Some rec. them with only one non-varying *H, but in most cases this is impossible in standard theory. Some have given ev. like *prH3-mo-> *-wo-(if H3m > H3w); whether these are regular & correct is not certain, but fits with *pro:-\ *pra:-varying. For most cases, more direct ev. is used. In part : *pet-'fall / fly' (or *pHet-if from met., see Ar. p'etur 'feather') *petH2-'fall / fly' *petH3-'fall', *ptoH3-mn *ptaH2k(h)-> G. πτήσσω \ ptḗssō 'crouch / cower for fear', Ar. t`ak`-čim,'hide oneself' *ptoH2k(h)-> G. πτώσσω \ ptṓssō, πτώξ 'cowering (animal), hare',-kh-(with other variants, some likely met. *petH2k-> *petak(h)-> *patek(h)-, *ptoek-, etc. (certainly *-k-with *pet(H)-) *pro *proH (most w/o distinguishing H2/3) *proH2, Dor. πρᾶτος \ prâtos 'first', *prH2-ei > *-ai ? *proH3, G. πρῶτος \ prôtos 'first', *prH3-mo-> *-wo-? *tu-'swell', *tuH-, *tuH1-, *tuH2-, *tuH3-(below) *krH3-'burn', Li. kùrti 'to kindle' *krH2-(a: > o: in Li. krósnis f. 'oven / kiln'), if related *wlHnaH2 'wool' > L. lāna, OIr olann, Go. wulla, *wilHnā > Po. wełna, Li. vìlna 'wool strand' *welH3no-> *wH3olno-> G. oûlos 'woolly / twisted / twined / curly / crinkled' *welH3->> *welH3mn > Arm. gełmn 'fleece' *wloH3mn > G. lôma 'hem / fringe' *wloH3k^o-> OE wlóh 'fiber / fringe', Ic. ló 'flock' (more ev. in https://www.academia.edu/126857503) *H2merg^-> G. amérgō 'pluck / pull' *H3merg^-> G. omórgnūmi 'wipe' *stlH2-*str(H3)-'stretch / expand' I said about one group in https://www.academia.edu/127198281 : > Knowing this, these names can be useful in analyzing other IE words. L. Mutunus Tūtūnus \ Mutinus Tītīnus 'phallic god of marriage' probably had dissimilation of ū-ū > *ū-ī / *ī-ū / ī-ī, since it came from *tuHto-\ *tutHo-> OIr toth 'vulva/vagina / fem. gender in grammar', G. sáthē 'penis', Arm. tutn \ ttun 'tail / end' (since many IE words can mean both: G. kérkos 'tail of a beast / penis'; more below). Note that G. sáthē 'penis' has the fem. gender in grammar, OIr toth 'vulva/vagina / fem. gender in grammar' has the masc. gender in grammar (old o-stem). Some IE words were used first for male genitals, then any genitals, then only female genitals (Skt. grdá-'penis', sárdi-grYdi-'vagina'; dual sakthyáu 'pole / shafts of a cart / vagina'; G. baubṓ(n) 'vagina / dildo'; sélīnon 'celery / *stalk > *penis > vagina'; Bq. potro 'testicle', potorro 'vulva'; Sino-Tibetan *puta 'penis shaft / vagina'). Arm. tutn \ ttun might retain *tH > t or be based on analogy with *tit 'breast' (merka-tit 'with bare breast(s)', titan 'a nurse', Luwian titan-'breast', OE titt). G. sáthē would show *tuH2to-> *twaH2to-> *tswatH2o-, however, this is disputed. In words for 'swell / be swollen/strong/firm', PIE seems to have *tuH3-, *tuH2-, tu-. In others, G. has tū-, which would (if all regular) come from *tuH1-: *tuH3lo-> G. sōlḗn 'channel/gutter/pipe/penis' *tu(H2)lo-> OE þol 'peg', G. túlos 'knot/callus/bolt', Skt. tū _ la-'tuft / wisp of grass / panicle of flower' *turo-> Skt. turá-'strong/abundant', turī _ pa-'semen' *tuHro-> L. ob-tūrāre 'stuff / fill up', LB tu-rjo, G. tūrós 'cheese', Av. tūiri-'milk that has become like cheese' *tuH3ro-> G. sōrós 'heap (of corn) / quantity' *tuH3ko-> G. sôkos 'bold/stout/strong one' *tuHko-> Slavic *tūkū > *tyky 'pumpkin', Greek tûkon / sûkon >> *t^ü:kos > *thü:kos > L fīcus 'fig', Arm. *thüg > t`uz There is a simple explanation for this. If H2 = x or χ and H3 = xW or χW, then dissimilation of *uxW > *ux could produce *tuH3-, *tuH2-. Depending on the nature of H1, a similar dissimilation might have given G. tū-, but direct *tuxW > *tuw > tu / tū is also possible (IE *H3 / *w also alternated). Supporting this is *tuH(2)turo-> G. Sáturos, Dor. Tī _ turos 'satyr / goat' with the same *u-u vs. i-u seen in Tutunus / Titinus. This is based on Solmsen's idea that it meant 'having a swollen penis' due to their nature : > Since the other roots with H2/3 have w or P, I think H3 = xW could dsm. > x or x^ near u \ w \ P (likely also KW). For *stlH2 \ *strH3, these are not usually seen as variants and do not fit. However, other types of dsm. exist. I see alt. r \ l in roots as caused by *H > *R; from https://www.academia.edu/129161176 : > Armenian & Tocharian show a lot of unexplained changes. There are many examples of PIE *r > l and *l > r. Though PT *ml-> TB ml-is regular, there is also *mluw-> TB pälw-. Maybe the same in Arm. *mlo:-> *pru:-> pr-(the regular outcome of *ml-is not known). Some words contain both oddities. ... Since almost all these words had PIE *H (2) or Ir. *h in them (mrāha-, maskah), it is likely that *H was
> pils-(vt.) '± stretch, strain [the ears]'PP /pepilso-/ klautsne=naiśai pepīltsoṣ śau[l]mpa m... more > pils-(vt.) '± stretch, strain [the ears]'PP /pepilso-/ klautsne=naiśai pepīltsoṣ śau[l]mpa mā spänteträ '[those who have] listened attentively, do not trust in life!' (3b4/5). ∎If correctly identified semantically, TchB pils-may reflect a PTch *päls-from PIE *pels-'quiver, tremble' otherwise seen only in Slavic [: OCS plachъ 'quivering, nervous, anxious' (< *polso-), plašiti 'be fearful' (P:801)]. In Slavic we have a semantic development *'quiver' > 'quiver with fear' while in Tocharian we see *'quiver' > 'quiver with excitement' vel sim. See also klausa-pilṣi. > With this, maybe : TB klau(t)sa-pilṣi 'with ears quivering, listening?, attent?' = S. upaśrutika-Now, if it was really *pels-> pils-, it would be in keeping with changes known from TB, though they don't seem regular: pils-from PT *piäls-from PIE *pels-. The changes to *iä & ä by P can include ä, u, i. However, consider what it would mean if from PIE *pyels-. Since IE 'mouse' & 'rabbit' are often < 'fearful / shaking', like : *petH2-'fall / fly', *ptoH2k-\ *ptaH2k(h)-> G. πτώσσω \ ptṓssō, πτήσσω \ ptḗssō 'crouch / cower for fear', πτώξ 'cowering (animal), hare', Ar. t`ak`-čim,'hide oneself' it allows *pyels-to have 0-grade *pils-: *pilsko-'mouse' > Sl. *pĭlxkŭ > R. poloxók, Po. pilch 'dormouse' The unique *-lxk-would be explained from *-lsk-. Since Celtic *il & *ir sometimes > ul & ur by P, likely also : *pilsko-s 'mouse / shrew' > *plikos-s > *plukot-s with *ss > *ts like S. Gaulish Lucotius, etc., favor *-ot-to *-u:t-given in https://en.wiktionary.org/wiki/Reconstruction:Proto-Celtic/luk%C5%ABts This would resemble another root (maybe with s \ H, ex. in https://www.academia.edu/128052798) : *py(e)lH1-> ON felmta 'be frightened / tremble', G. pállō 'shake/brandish', ptólemos / pólemos 'war' *py(e)lH1-? (if 'shaking / raging') > G. ptélas 'wild boar' The existence of *-i-& *-y-in both can hardly be unrelated. PIE *p-sometimes appears as Greek p-/ pt-/ ps-. Hamp said that this resulted from false division of *d#p > *t#p > #tp-> pt-, etc. This is not likely when G. should never have had *tp-to begin with, let alone preferred to analyze them in the exact opposite way expected. This sort of thing is known from E. (a-n-apron), but is most likely when a common word has 2 forms (*ainaz > a / an), allowing false division to create a reasonable alternative interpretation in speakers' minds. If from false division of *t#p, why would G. not also have many *k-> kt-for the same reason? Why also ps-from this? Why always followed by-Vl-? I said in https://www.academia.edu/128151755 that PIE had many *Cw-and *Cy-, and these are often seen in IE words, simply ignored. Even more of their effects are seen in alt., like *mwezg-, *muzg-. Why would *pels-form *pils-? It seems clear that a reevaluation of the standards of examining IE cognates is needed. Each piece of data that can not be explained with standard theory should not be ignored, but examined as ev. for possible theories and reconstructions closer to the truth.
The Greek god of war was Ares, but found in the name LB e-nwa-ri-jo \ e-nu-wa-ri-jo \ e-no-wa-ro,... more The Greek god of war was Ares, but found in the name LB e-nwa-ri-jo \ e-nu-wa-ri-jo \ e-no-wa-ro, G. Enualios \ Ἐνυάλιος, Lac. Inualios \ Ἰνυάλιος. These are from enualios 'warlike / fit for fighting / belligerent / crazy' (some of these might also be people named after the god or for hope for virtue in war). The name also in G. Ἐνυώ \ Enuṓ, a war-goddess.
A. The Philistines were said to have come from the land of Caphtor in the Bible, which has usuall... more A. The Philistines were said to have come from the land of Caphtor in the Bible, which has usually been seen as the same as Ak. Kaptaru & Egyptian Keftiw 'Crete' (*R > w known in Eg.). This was supported in the mid-nineteenth century by newly translated records from the region that also mentioned invading "Sea-People" with names that matched European locations. Later archeological finds showed that Mycenean pottery entered the region at the same time as the Philistines. So far, all evidence supported one conclusion. However, that idea was later challenged, in part based on the ideas of William Yewdale Adams that pottery and other elements of culture usually were introduced by trade, not invasion. His supporters had a major turnaround in academic circles, but in recent decades the invasion theory has gained support. Looking for DNA evidence a few years ago, more support for Cretan invasion appeared; see https://www.reddit.com/r/MinoanLang/comments/1hnekzq/philistines_messapians_crete/ So, how are Crete, Kaptaru, & Keftiw all related? Krētē must come from *Krwātā, *Kruwātā, or *Kruātā based on its legendary founder Krus (like Dôros founding the Dorians in myth, etc.). This would show *Kra:pta: in which-pt-> *-ft-> *-wt-> *-w-t-. For f > w, consider G. aleíphō 'anoint', *aleiwon 'oil' > *elai(w)on > élaion. Note that Crete was also called Khthonía '(is)land' (Κρήτη, Χθονία) from PIE *dhg^ho:m, *dhg^hom-'earth'. In Greek, some dia. had th > s, thus khth > ks in epíxenos 'upon the ground'. The Cretan Greeks also turned ks > *kx > *xR > *hR-in *ksew-'carve / scrape' > G. xū́ō 'scrape / scratch / shape by whittling/shaving / etc.', *ksustom > xustón 'spear/lance', Cretan rhustón 'spear'. Since Eg. Keftiw had its w < *R in standard thought, the need for *kx > *kR within Cretan is important. Other ex. of this change for *kR > k(h)r : Egyptian kekšer >> *kikhxor > *kikhror > G. kíkhora 'chickory' (r-r > 0-r) *kizdno-> *kistno-> *ksítanos > G. krítanos 'terebinth', Gmc. *kizna-> OE cén 'fir/pine/spruce' Khotanese kṣuṇa-'period of time, regnal period', Tumšuquese xšana-, *khs-> *khR-> G. khrónos 'time' *ksówano-'carving' > xóanon '(wooden) image/statue (of a god) / idol', *ksówano-> *kRówano-> Krónos Toxeús 'Archer', *Tokseus > *Teuksos > Teûkros (great archer from Salamis Island who fought in the Trojan War) Cretan shows words with alternation of m / p, and other Greek words do as well : *s(a)m-akis > Greek hápax 'once', Cretan hamákis Cretan kamá 'field', Dor. G. kâpos, Al. kopsht 'garden / orchard', ON hóf, OHG huoba, B. kapO / kOpO 'field / adjacent fields owned by same person' *kwa(H2)p-'foam / smoke / etc.' > G. kápnē \ kapnía 'smoke-hole' G. kámīnos 'oven/furnace/kiln/flue', NG kamináda 'chimney' *sH2aip-? > L. saepēs 'hedge/fence', G. haimasiā́ 'wall of dry stones' (if not <-*-mn-) G. hapalós 'soft / tender / gentle / raw (of fruit)', amalós 'soft / weak', Cretan hamádeon 'a kind of fig' (l > d like G. dískos, Perg. lískos 'discus/disk/dish', etc.) which seems like part of m / p / b in G. and other IE : *kH2am-> L. camur(us) 'bent', G. khamós 'crooked', khabós 'bent' G. kolúmbaina / kolúbdaina 'a kind of crab' (maybe a swimmer crab)
Piquero argued that the traditional source of Greek smáragdos 'emerald' (Semitic loan, cognate wi... more Piquero argued that the traditional source of Greek smáragdos 'emerald' (Semitic loan, cognate with Ak. barrāqtu, Heb. bāreqet 'emerald') is wrong, and that instead they are related to Eblaite wa-ru12-gatum, which he takes as a blue-green stone separate from emerald. He does so, in part, because (s)mfrom b-is problematic. Though it is, I can solve it in another way, that includes variants máragdos, zmáragdos, as well as barakís 'blue-green piece of clothing', and relates them to Linear B pa-ra-ku-ja 'emerald green? / inlaid with turquoise?' and similar words.
Melena considers Linear B *79 as WO2, representing wwo(:) and (*wyo(:) >) w^w^o(:). Though I agre... more Melena considers Linear B *79 as WO2, representing wwo(:) and (*wyo(:) >) w^w^o(:). Though I agree with most of his points, this can not be true since not all his words contain *ww from any source: *Diwós-sunos 'son of Zeus' > *Diwós-nusos > *Diwóh-nusos > Diṓnusos, with metathesis, also *Diwó(s)-nusos > Diónusos with *s-s > *0-s (or similar). Instead, they ALL have-wō-. From his evidence, there is no reason to take *79 as anything except WŌ. Though I agree with Melena's G. Maîa and LB ma-79 (hisma-wo2), and see more ev. for *-awya in Whalen 2025a, since it is from *Mawyōi, related to Maîa but with fem. *-o:y, this is a clear case in which *y-y > 0-y would be expected (it would be less likely for *-yo: to exist beside many other fem. *-o:y if one had to be chosen
Unfortunately, I've been too busy to individually work on each draft alone, but I felt having thi... more Unfortunately, I've been too busy to individually work on each draft alone, but I felt having this compilation of recent work on LA online in a place easily seen was needed. Each might be updated or fit together later. Some ideas have been updated in one, not the other.
Linear Algebra over Linear A? The signs used for numbers in Linear A, an ancient writing system f... more Linear Algebra over Linear A? The signs used for numbers in Linear A, an ancient writing system from Greece, are known because they are mostly simple dots & lines. Fractions are partly known, transliterated as A, B, C, etc., not fully known, but A is likely larger than B, B than C, etc. Some are certainly 1/2, 1/3, so a statistical approach was taken here: The mathematical values of fraction signs in the Linear A script: A computational, statistical and typological approach https://www.sciencedirect.com/science/article/pii/S0305440320301357
Melena (2022) examines occurrences of Linear B *86 in : me-86-ta 'an ethnonym' wa-86-re 'a shephe... more Melena (2022) examines occurrences of Linear B *86 in : me-86-ta 'an ethnonym' wa-86-re 'a shepherd's name' u-ra-86 'a place' ]86 'a place?' (same as me-86-?) This sign has been the most difficult to assign a value to, since only one word seems to be written out (in full, using separate signs of known value) another time in LB. Melena recognized it (1983), wa-86re must be one of the many PN's beginning with wa-duin LB and LA (including LB wa-du-na-ro, wadu-na, LA wa-du-ni-mi), often seen as "foreign" shepherds. Since LA names like pa-ja-re often appear
In Greek-like Elements in Linear A (https://www.academia.edu/58619465), Nagy provided an early an... more In Greek-like Elements in Linear A (https://www.academia.edu/58619465), Nagy provided an early and broad list and analysis of Linear A words matching Linear B, Greek, etc. For the 1st section, in 9., he considers that the names LA qa-qa-ru & LB qa-qa-ro might be related, maybe also G. Πάρπαρος. There is no certain IE ety., but maybe 'maker' < *kWr-kWr-<-*kW(e)r-(compare similar reduplicated Daedalus 'fashioner').This u \ o is important, and in others he did the same for LA ka-ru, LB ka-ro (G. Kallōn or Kha(i)rōn ?), LA pa-ra-tu, LB pa-ra-to (G. Platōn, or Palanthos 'bald' (words for 'bald' in names are fairly common)), LA ku-ru-ku, LB *ku-ru-ko, fem. kuru-ka (G. Glukos, Glukōn < gl(e)uk-'sweet'?), LA da-mi-nu, LB da-mi-ni-jo (G.-damno-or daimōn ?). Some with i \ e, maybe LA di-ki-se, LB de-ke-se-u, G. *Dexeus. Others with no changes (LA ma-ka-rite, G. Makaritēs), maybe showing that o > u and e > i were optional in some LA dia. (maybe applied to long & short, if *a: > *e: was unaffected).
For more, it refers to https://uralonet.nytud.hu/eintrag.cgi?locale=en_GB&id_eintrag=634 which only says that Lapp s- is secondary. How could they know? Why would it change *n- > sn-? I'd note that sn- appears in many IE words for 'snot', etc., notably *snudH- \ *snuHd- > Gmc *snutt- 'snot', *snu:t- 'blow the nose' (apparently related to W. nodd 'wetness / sap'). If a Gmc loan, it would be odd for sn- to only occur in Lapp and for it to be spread throughout Uralic.
Marek Stachowski in https://www.academia.edu/144024701 provides an alternative to 5 differentetym... more Marek Stachowski in https://www.academia.edu/144024701 provides an alternative to 5 differentetymologies that have been suggested for the Turkic word for 'garlic' :
In https://www.reddit.com/r/HistoricalLinguistics/comments/1n0czht/utoaztecan_w%C3%AF_o/ I said
t... more In https://www.reddit.com/r/HistoricalLinguistics/comments/1n0czht/utoaztecan_w%C3%AF_o/ I said that Uto-Aztecan *Cw & *Cy existed, partially seen by optional *yV > i, *wV > o \ u (no *e is known in PUA). Now, I have ev. that one root possessed both, seen by both changes in *syawha(l) \ *syahwa \ *syawa 'green / grass / leaf'. The movement of *h (or maybe older *x, etc.) is consistent with later variation of h & glottal stop. Indeed, with the "glottal stop hop" seen in the varying placement of glottal stops in clear cognates, I think *h or *x was was moved at an earlier stage. The changes of *syawhal > *siwi(l) are seen in *sawa ‘leaf' vs. *siwi(C) ‘green growth / green’. Stubbs : > 1294. *sawa ‘leaf’: VVH64 *sawa ‘leaf’; M67-255 *sawa ‘leaf’; B.Tep54 *haahaga ‘leaves’; L.Son233 *sawa ‘hoja’; CL.Azt97 *šVwV ‘leaf’; M88-sa1 ‘leaf’; Stubbs2003-45; KH/M06-sa1 *sawa: NP sawapi ‘sage’; TO; Nv; PYp; NT; ST; Eu; Tbr; Yq; My; Wr; Tr; Cr; Wc, CN. As one can see, a form of *sawa appears in every SUA language. Note Cr’s similarity to Tbr in *w > mw. [Tbr/Cr *w > mw] [SUA: Tep, Trn, Cah, Opn, Tbr, CrC, Azt] CN iswa-tl ... 1077. *siwi(C) ‘green growth’: AMR 1996d suggests *siwiC for Hp siwi ‘Parryela filifolia (shrub sp.) and CN siwi-tl ‘greenery, foliage, herb, leaf, turquoise, year’ as a separate set. Might this tie to *sawa ‘leaf’? [NUA: Hp; SUA: Azt] > His "Might this tie to *sawa ‘leaf’?" is essentially certain. It would be foolish to separate words varying only by *a vs. *i when alternations of V's are known within many other roots (those of exactly the same meaning). This can also be seen in alt. of *w \ *m (more below) in related *syahwal > *sa(h)mul : > 1057a. *(pa)-samaC / *-samhuC ‘grass’: BH.Cup *samVt ‘grass’; M67-204 *(pa-)sa/*(pa-)ca ‘grass’; CL.Azt237; Fowler83; M88-sa22; Munro.Cup53; KH.NUA; KH/M06-pa39: CL.Azt237 also discuss the difficulties of these words: Ca sámat ‘brush, herb, grass’; Cp sámat ‘grass sp.’; Ls şáámu-t ‘grass, hay, weeds’; Sr haamt ‘grass’; Ktn hamat; Sh sihmu ‘bunch grass’ matches Ls with i resulting from schwa-like behavior in the first vowel, and perhaps CN icmoliini ‘sprout again, grow, appear’ in the first two syllables, but not count yet. [NUA: Tak, Num] 1057b. *(pa)-soho ‘grass’ (< *-samhuC?): Hp söhö ‘galleta grass’; Hp(S) pashö; My básso ‘zacate’; AYq vaso ‘grass’. [NUA: Hp; SUA: Cah] > Here, there is clear ev. of *sam(h)aC vs. *soho within Root 1057. Since *wa > *o fits previous ev., older *sahwa > *soho is needed if they're related at all. Since both this & *sahwal > *sam(h)uC \ -aC are also needed, no simpler sequence exists. More ev. exists in the likelihood of *kw (very common in the world), with an environmentally iregular *Cw > *Cm (*Cw also seen in effects on V ) : *kwimya 'come' > NPt. kimma, Mn. kima [*my > *m(m) ] *kwimya 'come' > *kwomya ( > *kyomwa ?) > Tr. komu *kwimya 'come' > *kwiwya > *kwiya [dsm.] > N. ki/kiiwi 'come to do s.th.', SPt. -ki- 'come in order to’ *kwimya 'come' > *kmimya; likely *km- > *gm- > g- in Stubbs' "*We may want to keep in mind NT gíími 'ven acá!' and NT giíñ-kiaá 'ven acá!' in case the voicing in Tep is someday explained." The need for *-my- is probably also seen in Tübatulabal. There are 2 kinds of verbs with kVN-, and only one turns kVN- -> *VN-kVN- > kVŋgVN. Based on similar tïŋwa -> atelic *ïŋ-tïŋwa > ïndïŋwa 'to summon', it seems that only nasals in the same syllable were reduplicated. Thus, the alternation in Tü. *kwimya > atelic kima-t vs. telic *im-kim > iŋgim shows the presence of a nasal in the syllable. For *kwim-ya vs. standard *ki-ma, unexplained -mm- in cognates also favors *-my-. These alternations are known, but previously assumed to show simple irregular nasal assimilation (Stubbs) creating two classes of verbs with k-N, one in which an "excrescent N appears". However, if there was regular reduplication of nasals based on its placement in the syllable structure, then some of these are from older *kVN-(?)V, and no irregularity is needed
There are many similarities between Uralic languages and PIE, including the reconstructed PIElexi... more There are many similarities between Uralic languages and PIE, including the reconstructed PIElexicon. Many of these have been considered loans before. Though it's certain that some loans of existed, even
Francis-Ratte had regular *rC > *nC in PJ, but he also gave *ur-si > usi : > LAMENTS: MK wu:l-'cr... more Francis-Ratte had regular *rC > *nC in PJ, but he also gave *ur-si > usi : > LAMENTS: MK wu:l-'cries' ~ OJ urepe-'laments,' urayam-'envies,' u-si 'lamentable,' ura 'heart, mind'. pKJ *ur-'laments'.
PIE *tep-'warm / hot' also appears as *t(e)p(e)H1-in L. tepēre, *tpeH1-sk^e-> Al. ftoh 'cool'. IE... more PIE *tep-'warm / hot' also appears as *t(e)p(e)H1-in L. tepēre, *tpeH1-sk^e-> Al. ftoh 'cool'. IE *tp-> *pt-> ft-is probably reg. in Al.; other ex. come from met. of *t-p-> *tp-that could be late (https://www.academia.edu/143644895). There is other ev. that *-H1-existed in all forms. S. tápas-'heat' exists beside tápus+ (tápur-vadha-'striking like fire?', tápuṣa-aśna-'fiercely burning?') in a similar way to *g^en(H1)(o)s-'birth / family' > IIr. *j^anas-\ *j^anus-. Though most *H > i(:), some > u(:) (maybe in a similar way to *r(H) > i(:)r \ u(:)r). H-met. (https://www.academia.edu/127283240) may be the cause, since *tpeH1-sk^e-looks fairly odd, and *tepH1-sk^e-> Av. tafs-'become hot' is possible. There is more ev. in *tepH1to-> L. tepidus (with pre-aspiration https://www.reddit.com/r/etymology/comments/zuprzr/jens_elmeg%C3%A5rd_rasmussen/), since some IIr. point to *tH1epto-. Turner : > 5679 taptá 'heated, hot' RV. [√tap] Pa. tatta-, °aka-'hot, burning', Pk. tatta-, Gy. pal. tátă, arm. tatav, eur. tato, Ḍ. tʌ ́ta, f. °ti, Tir. táta, Woṭ. tat, f. tyet, Bshk. tʌtt, f. tεtt, Phal. tāto, f. °ti, Sh.gil. tātṷ, koh. tăto, gur. tāttŭ, K. totᵘ, S. tato (pp. of tapaṇu), L. tattā (pp. of tappaṇ), P. tattā, WPah.bhad. tattū, pāḍ. tāttā, cur. tattā, Ku. tāto, gng. tāt, N. tāto, B. tāt, Or. tatā, OAw. tātā, lakh. tāt, H. tāt, tātā, tattā, Marw. tāto, G. tātũ.-Verbal pres. stems formed from past participle: K. tatun 'to become hot'; Ku. tatoṇo 'to warm'; N. tātnu, tātinu 'to be hot, be energetic', tatāunu 'to warm'; B. tātā 'to be heated', tātāna 'to heat', Or. tātibā, tāteibā; Bhoj. tātal 'to be hot'; H. tatānā 'to heat'; Ext.-ll-: B. tātal 'hot, Or. tātalā, °tilā, Bhoj. H. tātal. Addenda: taptá-: WPah.kṭg. (kc.) tattɔ 'heated, hot', J. tātā, Garh. tātū. > There is also : S. taptá-'heated/hot/molten,' Ti. tath, A. táatu 'hot', Sh.d. tʌ ́to 'hot [of heated objects]', čhʌt 'hot [of the sun]', Indus Kohistani tʌ ̀th In some Dardic-th-, met. of aspiration from *thatta-is likely (seen in other words cognate with S. C-Ch-, etc.). If H1 = x^ (https://www.academia.edu/115369292), *tx^-> *th(y)-could give Sh. čhʌt. For other ex. of H1 \ y, see https://www.academia.edu/128170887. In others, older *th, even if not attested, is implied by tone. In A., other ex show alt of aa & *áa > óo. Retained áa must come from an older more complex tone; other Dardic clearly show that h & Ch caused low tone, so *thàptá > *thàptȃ (downstep) > *thàáptà (move tone back, one tone per mora) > *thăáptà (or a similar path). This also could be helpful in reconstructing other words not secure as cognates. In https://www.academia.edu/128170887 : PIE *tep-'warm, hot', PSCc *ṭep-/ *ṭp-*teplo-> OCS teplo-, OGr t'pil-i 'warm', Gr. tbil-, Mg. t'ubu OGr t'pebis , Gr. tbeba 'warm oneself', Mg. o-t'ibuans , Sn. t'ebid A relation with IE & SCc seems clear (for many other words), but here some have objected that-p->p-but t-> t'-makes no sense. If really *tH-, it would fit. Since H-met. happened within IE, often different in each branch, this would work best if SCc were a branch of IE.
A. The 2 nearly identical roots : *k^esH2-> G. keá-z-'split', eu-kéatos 'easy to split', Celtic *... more A. The 2 nearly identical roots : *k^esH2-> G. keá-z-'split', eu-kéatos 'easy to split', Celtic *kesta: > MI cess 'spear / javelin' *k^H2as-> S. śáśati 'cut', OCS kosa 'sickle' are likely the same, with H-met. in https://www.academia.edu/127283240 B. G. has several cases of mn \ *md > bd, often next to u : *tumdaros > G. Túndaros, Tundáreos, LB *tumdaros, *tubdaros, tu-da-ra, tu-ma-da-ro, tu-pa3-da-ro G. kolúmbaina / kolúbdaina 'a kind of crab' (maybe a swimmer crab) *wra(H2)d-> rhádamnos 'branch', rhámnos 'box-thorn', rhábdos 'rod (for punishment) / staff (of office) / wand' I think this is part of a larger alternation of P \ T near labials. In part https://www.academia.edu/120561087/Greek_and_Skt_P_dissimilation : *graphma > G. grámma, Dor. gráthma, Aeo. groppa 'drawing / letter' *samH2dho-> *(p)sam(a)tho-> G. fem. ámathos 'sand', psámathos, *psáthmos > *psáfmos > psámmos 'sand', *psámfos > Dor. psâphos 'pebble' There are plenty of ex. of certain IE origin, so Ranko Matasović's doubts in https://www.academia.edu/1489376 about *krumbo-& *krumdi-being IE are unneeded. It looks like another ex. of umb \ umd, just as likely as those in Greek. Adding to data & ideas rrom https://www.academia.edu/129170239 : *kH2awmb-> *kRamb-> 'wrinkled / shriveled' > G. krámbē 'cabbage', krambaléos 'dry', krombóō 'roast' *kRumb-> OE hrympel 'wrinkle', E. Shetl. krump 'crooked back', Sw. krympa 'shrink', ON kryppa 'hump / hunch', kroppr 'a hump on any part of the body', OI cromm, OBr crum 'hunchback', Br. kromm 'crooked' *krumbi-> *krumdi-> OI cruind 'round / globular / circular', OW crunn G. kramb-vs. kromb-might be from a > o near P (*madh-ye-> G. masáomai \ mossúnō 'chew'; G. ablábeia, Cr. ablopia 'freedom from harm/punishment'; *kapmos 'harbor' > Kommós; G. spérma 'seed', LB *spermo; *graph-mn > G. grámma, Aeo. groppa; *paH2-mn 'protection' > G. pôma 'lid / cover'; lúkapsos / lúkopsos 'viper's herb'; (a)sphálax / (a)spálax / skálops 'mole'; kábax 'crafty/knavish', kóbaktra p. 'knavery'; *H2merg^-> G. amérgō 'pluck / pull', omórgnūmi 'wipe'). Since Gmc used ablaut of a \ i \ u productively, *hrimp-doesn't require IE *kremb-, instead it could be analogy. C. PIE *(s)kr̥ dhu-has some oddities. G. σκυρθάλιος seems to show *r̥ > ur. This, of course, led Beekes to say, "The variation may point to a Pre-Greek word." However, the same in Indic https://dsal.uchicago.edu/cgi-bin/app/soas_query.py?qs=krdh : 3254 *kuḍha 'defective'. [See list s.v. kuṇṭha-] Pk. kuḍhiya-'blunt', m. 'fool'. Addenda: *kuḍha-[Perh. kr̥ dhú 'small, deficient' RV., Ir. *kr̥ du-in Shgh. čūrδ 'crooked', Pers. kul, Yazgh. ḱiδ] Shughni shows *kirdu-> *cirdu > *curdu, so it is likely that older *-dhw-could influence the V of *r̥ > *Vr. It's also possible, since Armenian shows *-ur-\ *-un->-r \-un-in some u-stems, that older *skr̥ dhur-existed, with some IE having met. or asm. > *skurdhr̥-or *skurdhur-. Though many IE roots with *(s)C-existed, Beekes wondered if G. σκυρθάλιος, Lac. κυρσάνιος vs. σκυρθάνια p. could be dissimilation of s-s after th > s. If so, it would make Doric th > s fairly old, which is also what I favor, even witih ex. from LA.
Uploads
Papers by Sean Whalen
compilation of recent work on LA online in a place easily seen was needed. Each might be updated or
fit together later. Some ideas have been updated in one, not the other.
*(s)nolke 'snot / saliva' https://en.wiktionary.org/wiki/Reconstruction:Proto-Uralic/ńolke :
Northern Sami: snuolga
For more, it refers to https://uralonet.nytud.hu/eintrag.cgi?locale=en_GB&id_eintrag=634 which only
says that Lapp s- is secondary. How could they know? Why would it change *n- > sn-? I'd note that
sn- appears in many IE words for 'snot', etc., notably *snudH- \ *snuHd- > Gmc *snutt- 'snot', *snu:t- 'blow the nose' (apparently related to W. nodd 'wetness / sap'). If a Gmc loan, it would be odd for sn- to
only occur in Lapp and for it to be spread throughout Uralic.
Based on other alt. of IE *d, *l with PU *δ, *l (below, F), I think that PU words with *-ke & *-
ka often being suffixes supports *snud-ke > *(s)nolke. Where the *-
H- appears, or its effects, is hard to say since it seems to disappear or move in IE.
For Hungarian nyál, in https://www.academia.edu/31352467 Zhivlov shows that *n could > ny near
*K. For apparent exceptions (his *n > ny out of nowhere), I said
in https://www.academia.edu/129090627 that *Kn- > ny- also. Since
in https://www.academia.edu/129640859 I said Turkic *kulxāk ‘ear’ ( Karakhanid qulaq, qulqaq,
qulxaq, qulɣaq ) and Uralic *kuxle- ‘hear’ (F. kuule-, Mi. kōl-, NMi. hūl-, etc.) were related and < PIE
*k^leus- 'hear'. With *-s- > *-x-, it could be that *sn- > sn- \ *xn- > ny- also. Without these changes,
both sn- and ny- would be unexplained.
that Uto-Aztecan *Cw & *Cy existed, partially seen by optional *yV > i, *wV > o \ u (no *e is known
in PUA). Now, I have ev. that one root possessed both, seen by both changes in *syawha(l) \ *syahwa \
*syawa 'green / grass / leaf'. The movement of *h (or maybe older *x, etc.) is consistent with later
variation of h & glottal stop. Indeed, with the "glottal stop hop" seen in the varying placement of
glottal stops in clear cognates, I think *h or *x was was moved at an earlier stage.
The changes of *syawhal > *siwi(l) are seen in *sawa ‘leaf' vs. *siwi(C) ‘green growth /
green’. Stubbs :
>
1294. *sawa ‘leaf’: VVH64 *sawa ‘leaf’; M67-255 *sawa ‘leaf’; B.Tep54 *haahaga ‘leaves’;
L.Son233 *sawa ‘hoja’; CL.Azt97 *šVwV ‘leaf’; M88-sa1 ‘leaf’; Stubbs2003-45; KH/M06-sa1 *sawa:
NP sawapi ‘sage’; TO; Nv; PYp; NT; ST; Eu; Tbr; Yq; My; Wr; Tr; Cr; Wc, CN. As one can see, a form
of *sawa appears in every SUA language. Note Cr’s similarity to Tbr in *w > mw. [Tbr/Cr *w > mw]
[SUA: Tep, Trn, Cah, Opn, Tbr, CrC, Azt]
CN iswa-tl
...
1077. *siwi(C) ‘green growth’: AMR 1996d suggests *siwiC for Hp siwi ‘Parryela filifolia (shrub sp.)
and CN siwi-tl ‘greenery, foliage, herb, leaf, turquoise, year’ as a separate set. Might this tie to *sawa
‘leaf’? [NUA: Hp; SUA: Azt]
>
His "Might this tie to *sawa ‘leaf’?" is essentially certain. It would be foolish to separate words
varying only by *a vs. *i when alternations of V's are known within many other roots (those of exactly
the same meaning). This can also be seen in alt. of *w \ *m (more below) in related *syahwal >
*sa(h)mul :
>
1057a. *(pa)-samaC / *-samhuC ‘grass’: BH.Cup *samVt ‘grass’; M67-204 *(pa-)sa/*(pa-)ca ‘grass’;
CL.Azt237; Fowler83; M88-sa22; Munro.Cup53; KH.NUA; KH/M06-pa39: CL.Azt237 also discuss
the difficulties of these words: Ca sámat ‘brush, herb, grass’; Cp sámat ‘grass sp.’; Ls şáámu-t ‘grass,
hay, weeds’; Sr haamt ‘grass’; Ktn hamat; Sh sihmu ‘bunch grass’ matches Ls with i resulting from
schwa-like behavior in the first vowel, and perhaps CN icmoliini ‘sprout again, grow, appear’ in the
first two syllables, but not count yet. [NUA: Tak, Num]
1057b. *(pa)-soho ‘grass’ (< *-samhuC?): Hp söhö ‘galleta grass’; Hp(S) pashö; My básso ‘zacate’;
AYq vaso ‘grass’. [NUA: Hp; SUA: Cah]
>
Here, there is clear ev. of *sam(h)aC vs. *soho within Root 1057. Since *wa > *o fits previous ev.,
older *sahwa > *soho is needed if they're related at all. Since both this & *sahwal > *sam(h)uC \ -aC
are also needed, no simpler sequence exists. More ev. exists in the likelihood of *kw (very common in
the world), with an environmentally iregular *Cw > *Cm (*Cw also seen in effects on V ) :
*kwimya 'come' > NPt. kimma, Mn. kima [*my > *m(m) ]
*kwimya 'come' > *kwomya ( > *kyomwa ?) > Tr. komu
*kwimya 'come' > *kwiwya > *kwiya [dsm.] > N. ki/kiiwi 'come to do s.th.', SPt. -ki- 'come in order
to’
*kwimya 'come' > *kmimya; likely *km- > *gm- > g- in Stubbs' "*We may want to keep in mind NT
gíími 'ven acá!' and NT giíñ-kiaá 'ven acá!' in case the voicing in Tep is someday explained."
The need for *-my- is probably also seen in Tübatulabal. There are 2 kinds of verbs with kVN-, and
only one turns kVN- -> *VN-kVN- > kVŋgVN. Based on similar tïŋwa -> atelic *ïŋ-tïŋwa > ïndïŋwa
'to summon', it seems that only nasals in the same syllable were reduplicated. Thus, the alternation in
Tü. *kwimya > atelic kima-t vs. telic *im-kim > iŋgim shows the presence of a nasal in the
syllable. For *kwim-ya vs. standard *ki-ma, unexplained -mm- in cognates also favors *-my-.
These alternations are known, but previously assumed to show simple irregular nasal assimilation
(Stubbs) creating two classes of verbs with k-N, one in which an "excrescent N appears". However, if
there was regular reduplication of nasals based on its placement in the syllable structure, then some of
these are from older *kVN-(?)V, and no irregularity is needed