¹ That is why I find rather strange what Bàrtoli says in Dalm., ii, 377. Of course “Tuscan” influence cannot be admitted for obvious historical, geographic and linguistic reasons, nor even “Venetian” influence, because Venetian palatalizes c(i) in ts or z (voiced s), whereas Dalmatian has č (= tš); but Bàrtoli seems to forget that once the whole of the palatalizing Romania had tš, preserved now (at least in part) in peninsular Italian, Corsican, Sicilian and Campidanese. For French this is proved by the preservation of č in the Northern dialects (cf. Meyer-Lübke, Hist. Gramm. der franz. Sprache,^4-5 pp. 124 f.; we know at all events that the č-area was once greater than it is now); for Spain, by the Arabic transcription of Old Spanish names (Elche) and words, and by some relics of č in Modern Spanish (cinche > cīmicem, chícharo >∗ cicerum; cf. Meyer-Lübke, RFE, vii (1921), 227 fi.; 250; ibid., xi (1924), 23 ff.; Menéndez Pidal, Manual,⁶ pp. 120 f.; A. Castro, RFE, i (1914), 102; Lapesa, Historia de la lengua española (Madrid-Buenos Aires-Cadiz, 1942), pp. 47; 71; 74; 243; and Bàrtoli himself in Miscellanea Hortis, Trieste (1910), p. 915, n. 1. Bàrtoli seems to have modified — and improved — his point of view in RDR, ii (1910), 461 ff. Cf. also Friedwagner, ZRPh, liv (1934), 677, n. 7; Krüger, Lbl. f. germ. u. rom. Phil. xxxix (1918), 125.

² Of course, the old neo-grammarian school would reason (and in effect did reason) like this: in Vegliotic ci palatalizes, but ce does not; but in the other continental Western Romance languages both ce and ci palatalize; the two phenomena are different, and therefore independent of each other. I have criticized this methodological error in RIGI, xv (1931), 163 ff. (Come si estende il processo fonetico), and see no need for repeating here my arguments, especially since the fact would by now be commonplace for many linguists.

³ Romanian also shows the palatalization of ce, ci, ge, gi, which some scholars consider to be independent of the Western one (so e.g. Skok, ZRPh, xlvi [1926], 409; l [1930], 511; Meyer-Lübke, Mitt. des rum. Institutes, i [1914], 13). I do not need to go into this problem, however, first, because even if there is a connection, and if the Romanian palatalization is of Western origin, there is no proof whatever that it passed through Dalmatia (it could have been brought into Dacia directly by the Roman colonists, as in fact I believe); second, because a close dialectal grouping of Romanian and Dalmatian has to be discarded, in my opinion (the argument furnished by kopsa etc. has been destroyed by Graur and Rosetti, Bull. Ling., vii [1939], 185). All the isoglosses of Dalmatian (except kopsa, now eliminated) point towards the West, not the East. But of course, for the theoretical point I am making here, it is quite immaterial whether the palatalizing wave reached Dalmatian from the West or from the East, for Romanian palatalizes c (and g) in exactly the same way in front of e and in front of i, just as the Western Romance languages.

Dalmatia was Romanized either after 155 b.c. (conquest) or after 11 a.d. (repression of the last great insurrection); Dacia, after 107 a.d.

⁴ Lat. c in Vegliotic palatalizes into č also before ü (< Lat. ū) and before (< Lat. ě, ae, as in Italian, French, Spanish, etc.): čol < cūlum, < nec-ūnum, sčor < secūrim, sečoira < ∗siccūra, sčor < obscūrum; čil < caelum (Ital. cielo, Fr. ciel, Sp. cielo), lačar < lacertum, čart < certum (Span. cierto), čant < centum (Spanish ciento); also -čal = Ital. -cęllo < Lat. -ěllum (munčal = Ital. monticęllo; basalčala; cf. Bàrtoli, Dalm., ii, pp. 377 f.). Both and ü have of course a much closer articulation than e (the opening of the mouth is not so great).

“Panromanic” ci (resulting from Latin qui, probably by dissimilation) also palatalizes: = Ital. cinque, Fr. cinq, Span, cinco; = Ital. cucina, French cuisine, Span. cocina.

⁵ The phonetic difference between the e-palatalization and the i-palatalization has been carefully studied by Olaf Brok (= Broch) in olděljenija russkago jazyka i slovjesnosti imperatorskoj akademij nauk, 83, n. 4, Sanktpetersburg, 1907 (Opisanije odnogo govora jugozapadnoj časti totemskogo uězda, pp. 75-91, §12, Momjenly palatalizacij) and by Roman Jakobson, Fonjetika odnogo sjevjerno-vjelikorusskogo govora s namječajusčjejsja pjerjexodnost'ju (Prague, 1927), pp. 57-62, §10. I owe this indication to the kindness of my good friend Roman Jakobson.

⁶ On this subject see in general Bonfante, I dialetti indoeuropei (Naples, 1931), pp. 131 ff., especially pp. 133 f. (with n. 1 p. 133); cf. also p. 153; Emerita, ii (1934), 186 f.; for Armenian see now Bonfante, Mélanges Pedersen, pp. 26 ff. (Armenian, very close to Greek, also shows a less extensive palatalization than the other languages; the epicenter of palatalization, as of many, if not of most, innovations of Indo-European times, was obviously in the Slavo-Iranian area, as I hope to prove soon in an extensive work). Cf. also the masterful remarks of Campus in Atti della accademia delle scienze di Torino, liv (1919), 110 (with more bibliography of the same). The old opinion is asserted e.g. by Meillet in Mém. soc. ling., viii (1894), 285: “La palatalisation grecque est indépendante de celle de l'indoiranien, du letto-slave [sic!] et de l'arménien.” Cf. also the same in Bull. soc. ling., xx (1916), 44.

⁷ Lithuanian also lies on the fringe of the palatalizing area, so the palatalization is very slight (k'e, k'i) and not even marked in writing (it is, I think, phonetic and not phonemic); cf. Leskien, Litauisches Lesebuch, p. 146.

⁸ Arcadic shows some complications, doubtless because of dialectal superpositions; cf. Bechtel, Griech. Dialekte, i, pp. 328 f.

⁹ Cyprian has the same opposition, only with σ instead of τ: , but , .

¹⁰ The same difference in the treatment of consonants in front of e and of i is to be found in Romanian, where (Proto-Romanian) ti, di, si, li, pi, bi, fi, vi, mi palatalize, but not (Proto-Romanian) te, de, se, le, pe, be, fe, ve, me: cf. totī < tōtī, suptire < subtīlem, capatîna < ∗capitīnam, but frate < frāter, te < tē, teme < tīmēre, < ∗tīium (cf. tīlia); zic < dīcō, cruzī < crūdī, but de < dē, deget < dīgīium, des < dēnsum; rânşinân < resīnam, leşie < lixīuam, deşī < dēnsī, şi < sīc, but se < sē, searân < sēra, semn < sīgnum (cf. ital. segno); in (Macedo-Romanian l'in) < līnum, i < (il)lī (cf. Ital. gli), but lege < lēgem, lin < ∗lēnum (cf. lēnis), ling < līngō, limbân < līnguam, etc. etc; cf. Tiktin, Elementarbuch (Heidelberg 1905), pp. 60 f.; 62; 65; 93; Densusianu, Histoire de la langue roumaine, i (Paris 1901), pp. 283; 307 ff.; M. Friedwagner, ZRPh, liv (1934), 670 ff.; Meyer-Lübke, Mitteilungen des rumänischen Institutes an der Universität Wien, i (1914), 13 ff.; Rosetti, Istoria limbii române, iii (1940), p. 41, with bibliographical indications, and now the beautiful Atlasul linguistic român (Cluj 1938) i, maps 46 (ficat), 73 (visez); 19 (pupilân, luminân), 40 (şira spinarii); 85 (suspin), 93 (pisc) etc., etc. and the Micul Atlas, maps 7b; 81; 30; 106; 107; 115; 132; 137; 172a.

The same is true for the non-Doric Greek dialects (Ionian and “Achaean”), where (non-initial) Indo-European ∗ti (unless preceded by s) assibilates (through palatalization) to si: Doric : Ionic , Arcadian , Lesbian ; Doric : Ionic etc. etc. In initial position the articulation of the consonant was stronger and ti remained: ; cf. Schwyzer, Griech. Gramm., pp. 270 f. for is rare and appears only in the last syllable () or in interior position before vowels (; unique and very doubtful).

In Hittite, Indo-European ti passes to zi (= tsi) unless s precedes (just as in Greek): asanzi = Sanskrit sánti, Greek etc.; but Indo-European te remains: sawetesz = Homeric Greek genit., Lat. ueteris genit.; esten = Greek , Latin estis, etc. Perhaps ∗tē also becomes ze: cf. zek (always written zik!) = Latin tē; in this case, we have to admit that ē was closer than ě, as in Vulgar Latin. Cf. Sturtevant, Compar. Grammar of Hittite (Philadelphia, 1933), p. 126; Bonfante, Revue belge de philologie et d'histoire, xviii (1939), 382.

In Vegliotic itself, tī is palatalized in final (unstressed) position, but not te: cf. Vegliotic diañč < dentī(s), tañč < tantī, aniñč < ∗inantī, veñč < uīgintī; in Friulian and in Lombard also lī, dī, nī are treated the same way: Milanese kavay, müy, fradey, ań, pañ, deñč, fañč, tüč etc.; cf. Meyer-Lübke, Gramm. rom., i, pp. 274 f.; Mitt. des rum. Inst. Wien, i (1914), 15f. The unstressed position seems to favor palatalization: Lesbian has always τε = Lat. que, where we would expect ∗πε cf. Bonfante, I dialetti indoeuropei, p. 135 n., and see what is said above, in this same note, about Greek .

¹¹ This can be easily confirmed now by the AIS.; cf. e.g. the maps 5 padre; 8 madre; 179 cacare, caca; 552 la sega a mano, 553 la sega lunga, 554 la segatura, 555 segare (Southern Italy has sek-); 567 il sugo dell'albero (Southern Italy has suk-); 622 ortica; 677 la diarrea (cacaiola); 799 il sagrestano (Southern Italy has often -kr-); 802 pregare; 949 asciugano (-g- only in Tuscany and Latium); 1046 vitello; 1227 la ruota; 1235 le redini (South It, has ret-); 1289 il fico, i fichi; 1589 pagatemi (pakat- in points 645; 710; 713; 749; 938; 937); 1614 pagati; 1236 la cavezza (South. Italy has kap-); 1459 spiga; 1460 spigolare (South It. has -k-); 1539 ago (all Southern Italy has ak-) etc., etc. It will be noticed that in pācāre (between two a) -c- sonorizes on a much wider area than in secāre (between e and a), sūcum (between two u), and spīcam (between ī and a) etc., etc. See now R. Hall in Language xix (1943), 127 ff.

I call “Southern Italy,” following the Italian usage, the whole of Abruzzi, Campania, Lucania, Apulia, Calabria, and Sicily.

I follow therefore, of course, the neolinguist Bàrtoli, in Meyer-Lübke, Grammatica storica della lingua italiana,² translated by Bàrtoli and Braun (Turin), p. 98. This work reflects Bàrtoli's, not Meyer-Lübke's ideas; cf. Italia dialettale xiii (1937), 227; Bàrtoli, Introduzione alla neolinguistica (Geneva, 1925), pp. 58 ff.; Studi albanesi, ii (1932), 26 ff. The neogrammarian Meyer-Lübke, Grammaire des langues romanes, i (Paris, 1890), 383 ff., Italienische Grammatik (Leipzig, 1890), pp. 115 ff.; 122 ff. of course solved the problem with the magic wand of the “phonetic law” (influence of the accent).

¹² As I do not believe (and, I think, nobody believes by now) in any influence of accent on the sonorization of the intervocalic stops, I would add padella, ladino, padire, gradire piagere, piagenza, staduto, madornale, badessa, badía, gradella, armadura, ciurmadore, mallevadore etc.; pagare, dragone, magello, vagellare, dugento, dragone, aguto, aguzzo, badile, caviglio, savore, savone, savere (Dante) etc.

That there should be some oscillation (piacere, acuto, sapone, statuto, latino, armatura, sapere, sapore, patire, macello, gratella, avvocato), in part under learned influence, is of course quite natural.

I omit words of sure French (aguglia, masnada, rugiada, congedo) or Venetian (lido, laguna, doge) origin, as well as obvious book-words, such as strato.

¹³ The influence of a following a is admitted by Meyer-Lübke, Grammaire des langues Romanes, i (Paris 1890), p. 383: “les explosives sourdes placées entre une voyelle accentuée et une atone persistent en roumain et dans l'italien du Sud; ailleurs elles deviennent sonores, dans l'italien du centre seulement devant un a [Italics mine], devant n'importe quelles voyelle dans tous les autres domaines.” If true, this fact would of course confirm my point. (Meyer-Lübke's accentological “rule” cannot be accepted, of course; see here notes 11 and 12.)

As a true neo-grammarian (although certainly one of great merits), Meyer-Lübke does not even consider the problem of the center and of the expansion of the innovation -g-, -d-, -b- (-v-), which is on the contrary essential and primary for neo-linguists like Bàrtoli and myself.

¹⁴ The same idea was already expressed by Von Wartburg, but in an abstract way, without any proof or example (not even for the problem he treats: ū > ü in Romance), in ZRPh, lxvi (1936), 14: “Dazu kommt, dasz die umgebenden Konsonanten zum Teil hemmen, zum Teil beschleunigen.” Much more important is Fr. Schürr, ZRPh, xlvi (1976), 296: “Dasselbe Beispiel mag dann auch noch zeigen, wie ein Lautwandel bei seiner räumlichen Ausbreitung einschränkenden Bedingungen unterliegen kann. Der Wandel a > e, der sicherlich von der Romagna ausgehend die Emilia erfaszt hat, erscheint im nordwestlich anschlieszenden Gebiete, in Piemont, nur vor r (so namentlich in den Infinitiven I) [—-]. —– Oder man mag daran denken, dasz der florentinisch-pistojesische Wandel von l ^kons > i sich jenseits des Appeninkammes nur mehr vor Labial und Velar findet (vgl. RDst., ii, S. 235), also eine Beschränkung auf ihn begünstigende phonetische Bedingungen erfährt.”