A unified account for the behaviour of Gallo-Romance glides in

Dec 5, 2009 - the key to the apparently anarchic behaviour of glides is this bleeding relationship: 1. the chronology of the events allows for a phonologically ...
129KB taille 0 téléchargements 208 vues
Philippe Ségéral Université Paris 7 [email protected]

Going Romance 23 Workshop on Linguistic Change in relation to Linguistic Theory

Tobias Scheer CNRS 6039, Université de Nice [email protected]

December 5th, 2009 Nice

this handout and some of the references quoted at www.unice.fr/dsl/tobias.htm

A unified account for the behaviour of Gallo-Romance glides in strong position (1)

purpose a. offer a unified account for the evolution of glides – yod and w – in Strong Position in Gallo-Romance: 1. strengthening to a contour segment (affricate, gw) occurred in all cases 2. except in case the input configuration – the strong position of the glide – was destroyed by independent processes before strengthening was active b. ==> the key to the apparently anarchic behaviour of glides is this bleeding relationship: 1. the chronology of the events allows for a phonologically exceptionless analysis. 2. any glide that was still present in Strong Position "late" undergoes strengthening. c. this analysis may also be extended to other sonorants if epenthesis is considered a strengthening: cam(e)ra > chambre where the br is a contour segment (which is the strong version of r).

(2)

roadmap a. yod in initial position b. yod in post-consonantal position c. w in initial position d. w in post-consonantal position

1. Yod in initial position: > dÉʒ without exception (3) lat.

jocu jurare

# __ jeu [ʒø] jurer [ʒyʁe]

coda intervocalic __ # V __ V maj(u) mai [mɛ] raja raie [rɛ] jejunu jeûn [jœn]

germ. *jok ofr joc (> jucher) *jehhjan ofr jehir, gehir, jeir blending with géhenne produces mod fr. gêner, cf. Rey (1998) géhenne, gêne

-2-

2. Yod in post-consonantal position (4)

origin: consonification of i,e in hiatus position fiilia > filja fille viinea > winja vigne

(5)

global picture a. (almost) exceptionless strenthening after labials (rabia > rage) b. a variety of evolutions elsewhere 1. strengthening liineu > linge 2. persistence of yod modiolu > ofr moiel 3. palatalization glacia > glace 4. metathesis baasiaare > baiser

(6)

important distinction: C+yod after vowels vs. after consonants e.g. gj a. always > yod in V__ regioone > ofr roion b. always > strengthening in C__ Georgiu > Georges

2.1. Labial + yod (7)

V__: strengthening bj > dÉʒ (> ʒ) vj > dÉʒ (> ʒ) rabia rage cavea cage rubeu rouge leviu liège goobioone goujon abbreviaare abréger

mj > ndÉʒ (>~ʒ) pj > tÉʃ (> ʃ) vindeemia vendange apiu ache siimiu singe sapiam sache *blasteemia a.fr. blastenge seepia seiche

(8)

C__: strengthening Cbj > dÉʒ (> ʒ) Cvj > dÉʒ (> ʒ) Cmj > ndÉʒ (>~ʒ) Cpj > tÉʃ (> ʃ) longe serviente sergent commeaatu congé *krippja crèche *lumbea cambiaare changer *cervia ofr. cerge superbia ofr. soverge *conserviu ofr. concerge

(9)

small paradigm where persistence is observed the list below is about exhaustive bj > dÉʒ (> ʒ) vj > dÉʒ (> ʒ) mj > ndÉʒ (>~ʒ) habeo ai *aviolu aïeul – – deebeo ofr. dei *gaveola ofr. jaiole *plovia pluie

pj > tÉʃ (> ʃ) sapio ofr. sai

-3(10) the yod is a geminate consensual: e.g. La Chaussée (1974: 67, 171), Bourciez & Bourciez (1967: §40, §49-H), Fouché (1966-73:256, 906, 909-R1) a. lab + yod > yod only occurs after vowels b. intervocalic yod regularly disappears: raja > raie c. the labial is lost, which indicates that it stood in coda position d. the evolution of the preceding tonic vowel follows the closed syllable paradigm: sapio > ofr. sai, habeo > ofr ai

2.2. Yod after non-labials solution 1: strengthening > dÉʒ, tÉʃ V__ tj – –

dj

sj kj gj nj

lj rj

C__ Ctj porti(c)u Perti(c)u Aventi(c)u excorti(c)at ttj *matteuuca Ctj hordeu vir(i)diaariu indeusque

gage siège piège ofr. miege ofr. forasche – ofr. domesche – Ckj *ankja *nuskja – – Cgj Georgiu spongia liineu linge mnj somniaare étrange somniu extraaneu laaneu lange *dom(i)nioone *fanja fange *dom(i)niaariu *mentioonia mensonge Catalauni(c)u Chalonge Santoni(c)u Saintonge mani(c)u ofr. mange alveu auge – salvia sauge *sturione esturgeon rrj *burrione ceereu ofr. cerge sorooriu ofr. serorge *camoria ofr. chamorge baleaari(c)u ofr. baillarge cleeri(c)u ofr. clerge

*wadiu *sedi(c)u *pedi(c)u medi(c)u forasti(c)u domesti(c)u –

porche Perche Avenche écorche massue orge verger ofr. enjosque – anche ofr. nosche Georges éponge songer songe donjon danger

– bourgeon

collier, bracelet

-4solution 2: persistence yod > yod V__ ofr. moiel – dj modiolu gaudia joie inoodiaare ennuyer gj regioone ofr. roion – exagiu essai corrigia courroie

C__ – –

solution 3: palatalization C+yod > tÉs (>s) V__ C__ tj ratioone raison Ctj cantioone chanson *latia laize fortia force *acutiaare aiguiser *linteolu linceul enfance infantia ktj *tractiare tracer factione façon lectione leçon suspectione ofr. sospeçon frictione frisson maledictione maudisson *districtia détresse kj glacia glace Ckj arcioone arçon suspicioone soupçon lancea lance *gloociaare glousser calceaare chausser nj viinea vigne Cnj hernia ofr. hergne oignon Arvernia Auvergne uunioone Burgundia Bourgogne *vernia vergne lj palea paille – – taaliaare tailler melioore meilleur solution 4: metathesis C+yod > IC V__ ssj sj baasiaare baiser rj

paria

paire

C__ *bassiaare baisser passione paisson – –

(> soupçon)

-52.3. Analysis (11) traditional analysis Pope (1952:§203), La Chaussée (1974:79), Carton (1974:162), Zink (1986:101), Jacobs (1993:149), Bourciez (1967:§171) a. yod is particularly vigorous and cannot stand being in this position b. it "tries to escape" by all means, which are: 1. destruction of the preceding consonant (= persistence): modiolu > ofr moiel 2. palatalization = it enters the preceding consonant 3. when neither is possible because of the nature of the preceding consonant, it crosses this consonant c. none of these strategies works with labials, which are particularly strong and impalatalizable. ==> strengthening is a last resort if all other means fail. d. strengthening after non-labials is the treatment of "late" or "retarded" words, i.e. 1. Germanic words that came in late: *wadiu > gage 2. words whose evolution was retarded due to extralinguistic reasons: learned, liturgical vocabulary: liineu > linge Bourciez (1967: §149-H, R1, §199-R3, §182-H) La Chaussée (1974: 183) after labials: end 3rd, beginning 4th AD La Chaussée (1974: 193) after other consonants: end 4th, beginning 5th AD (12) same input, different development exhaustive list of contexts other Vrj metathesis paria paire Ckj palatalization arcioone arçon Vdj persistence modiolu ofr. moiel Ctj palatalization cantioone chanson Vnj palatalization viinea vigne

strengthening *sturione esturgeon *ankja anche *wadiu gage porti(c)u porche liineu linge

(13) all cases of strengthening are "late" a. Germanic origin: *sturione, *ankja, *wadiu Germanic vocabulary systematically undergoes strengthening by affrication b. consonification had to "wait" for the loss of an intervocalic consonant -Cti(c)u porti(c)u > porche -Vt/di(c)u herbaati(c)u > herbage medi(c)u > ofr. miege c. extralinguistic reasons for inhibiting evolution: liturgical etc. vocabulary liineu > linge

-6(14) labials vs. other consonants a. labials resist longer in coda position than other consonants b. recall the contrast between cavea > cage and aviolu > aïeul c. cavea > cage is a strengthening in post-consonantal position: it supposes the PRESENCE of the labial at the time when strengthening takes place. d. aviolu > aïeul is a gemination of the post-consonantal yod on the position that is vacated by the labial. It supposes the ABSENCE of the labial at the time when strengthening takes place. σ σ σ σ σ σ | | | | | | R R R R R R | | | | | | O N C O N O N O N C O N O N | | | | | | | | | | | | x x x x x x > x x x x x x | | | | | | | | | | a v j o l u a j eu l e. f.

==> "persistence" is a form of strengthening: 1. the weak coda consonant is eliminated 2. the strong post-coda consonant spreads on its position two types of strengthening 1. by affrication: yod > dÉʒ, tÉʃ 2. positional: (compensatory) gemination j > jj

(15) same contrast with Vdj a. modiolu > ofr. moiel vs. *wadiu > gage b. same analysis: modiolu > ofr. moiel = loss of d in coda position, positional strengthening *wadiu > gage = strengthening by affrication in post-d position, THEN loss of d in coda position

-7(16) positional analysis [building on Ségéral & Scheer 2001] a. period 1 strengthening by affrication not active metathesis, palatalization and positional strengthening active b. which process C+yod clusters undergo depends on segmental factors: 1. only s,z,r allow for metathesis: paria > paire 2. other consonants do other things: d is lost in codas modiolu > ofr moiel etc. b. period 2 strengthening by affrication active metathesis (and palatalization, positional strengthening?) not active c. paria modiolu glacia *wadiu medi(c)u liineu period 1 – – – metath. paire – – – strength gem. mojjel – – – palataliz. glace – – – period 2 strength affric gage miege linge d. e.

all yods which were still present in post-coda position at the time when affricatestrengthening was active have undergone this process Yods that escaped affricate-strengthening 1. were either no yods anymore at that time 2. or did not stand in strong position after an independent coda consonant anymore ==> affricate-strengthening is bled by processes that have applied earlier

(17) comparison traditional analysis a. yod is particularly vigorous: it kills preceding consonants b.

explanation of the contrast labials vs. non-labials labials are strong: they are particularly resistant against palatalization

positional analysis yod is passive and opportunistic: it spreads on the position of preceding consonants once they die of age explanation of the contrast labials vs. nonlabials labials are strong: they are particularly resistant in coda position

2.4. Loss of the medial consonant in C[C]C (18) general process: loss of the medial consonant in a tri-consonantal cluster [if none of the neighbours can build a branching onset] -l[b]n- gal[b](i)nu jaune -l[k]t- cul[c](i)ta ofr. coute -m[p]t- com[p](u)taare conter -n[d]t- ven[d](i)ta vente ofr. dert[r]e -r[m]t- dor[m](i)tooriu dortoir -r[b]tder[b](i)ta -s[t]m- tes[t](i)mooniu témoin -r[k]b- ar[c](u)ballista arbalète -r[p]m- *car[p](i)mu charme -r[t]k- excor[t](i)cat écorche

-8(19) hence all Cj clusters after consonants reduce: C[C]j > Cj loss of labials loss of non-labials longe C[t]j por[t]i(c)u C[b]j *lum[b]ea C[p]j *krip[p]ja crèche C[d]j hor[d]eu C[v]j ser[v]iente sergent C[k]j *an[k]ja C[m]j com[m]eaatu congé C[g]j Geor[g]iu C[n]j som[n]iaare C[r]j *bur[r]ione

porche orge anche Georges songer bourgeon

(20) hence a. por[t]i(c)u > porche hor[d]eu > orge are NOT palatalizations of t,d, but affricate-strengthenings of a yod after an r por[t]i(c)u = *sturione b. voiceless tÉʃ > ʃ witnesses a chronology that goes through rç, rather than rj: porti(c)u > por[t]ju > por[t]çu > porçu > porche c. also attested: porge supposes rj, i.e. witnesses the loss of t, which occurs before it can devoice j into ç (21) reality of the loss of the medial C n[d]j > ¯ vereecun[d]ia vergogne *retun[d]iaare rogner Burgun[d]ia Bourgogne Compen[d]ia Compiègne

m[n]j > ~dÉʒ som[n]iaare som[n]iu *dom(i)[n]ioone *dom(i)niaariu

songer songe donjon danger

(22) benefits a. apparent palatalization of t into tÉʃ *foras[t]i(c)u > ofr. forasche (> farouche) domes[t]i(c)u > ofr. domesche 1. classically two distinct palatalizations need to be assumed depending on the palatal agent: t > tÉs /__i t > tÉʃ /__j 2. this is a mirage: the evolutions at hand concern sj, not tj: s[t]j > sj > stÉʃ (>tÉʃ > ʃ) 3. this tells us that metathesis was not active anymore when sj was created. b. the same goes for al[v]eu > auge sal[v]ia > sauge the consonne d'appui is the l, not the v c. the same goes for C-labial-yod ser[v]iente > sergent the consonne d'appui is the r, not the v

-9(23) metathesis was not active anymore in period 2 grey-shaded items have the relevant input for metathesis, but because the relevant input context was created too late. VCj CCj sj baasiaare baiser ssj *bassiaare sj foras[t]i(c)u ofr. forasche ttj *matteuuca rj paria paire Ctj por[t]i(c)u tj ratioone raison Ctj infan[t]ia

do not effect metathesis baisser massue porche enfance

(24) chronology a. 1. palatalization, metathesis (period 1) 2. reduction of triconsonantal clusters C[C]C (period 1) 3. affrication-strengthening (period 2) b. doublets (many more to be found in diatopic variation) palatalization affrication-strengthening Vnj *fanja fagne fange Vrj *camoria chamorge c.

metathesis chamoire

the original form of the word persists alongside with its evolved version, both being semantically or diatopically differentiated. By the time affrication-strengthening enters the scene, the non-evolved C+yod version then undergoes this process.

2.5. Palatalization: a confetti analysis (25) positional strengthening: gemination of post-coda yod a. gemination without coloration gemination with coloration Vpj sapio ofr. sai Vnj viinea vigne Vbj deebeo ofr. dei Vlj palea paille Vvj *aviolu aïeul Vkj glacia glace ofr. moiel Vdj modiolu ofr. roion Vgj regioone b.

all items at hand are geminates 1. the preceding tonic vowel shows the effects of closed syllables apj sapio ofr. sai agj glacia glace alj palea paille anj montanea montagne 2. the reflex of kj is tÉs, which resists intervocalic voicing: glacia > glace it thus behaves like ttj: *matteuuca > massue compare with singleton tj > tÉs that in addition voices: *latia > laize

- 10 (26) why does the coda consonant sometimes leave a (melodic) trace, but at other times leaves the geminate colourless? a. it cannot be argued that - coloured geminates are based on palatalizable consonants: n,l,k - while uncoloured geminates are based in non-palatalizable cons.: p,b,v,d,g b. rather, coloration depends on the general behaviour of each individual consonant in coda position: 1. labials and dentals are lost without leaving any trace: rupta > route cub(i)tu > coude naav(i)gaare > nager plat(a)nu > ofr. plane adveniire > avenir 2. velars are resolved into yod, but which of course cannot make the geminate yod more palatal than it is facta > faite rig(i)da > ofr roide 3. nasals nasalize leave a nasal trace on the preceding vowel cantaare > chanter in our case the nasality lands on the following consonant: viinea > vigne 4. the same goes for the lateral palea > paille the leftovers of l are w in codas, but laterality here - why? Because laterality cannot land on a vowel, and w cannot land on a yod. (27) the confetti analysis a. when a consonant is eliminated - either in coda position - or as the medial consonant of a triconsonantal cluster C[C]C pieces of its melody may hook on the remaining items b. viinea > vigne is not a palatalization of n, but a nasalization of jj palea > paille is not a palatalization of l, but a lateralization of jj consonification O N O N O N | | | | | p a l e a [le]

>

O N O N O N | | | | p a l j a [lj]

1. loss of coda l 2. gemination >

O N O N O N | | | | p a l j a

lateralization of the geminate >

O N O N O N | | | p a l+j a [ʎʎ]

- 11 consonification O N O N O N | | | | | gl a c i a [ki]

>

O N O N O N | | | | gl a c j a [kj]

1. loss of coda l 2. gemination >

O N O N O N | | | | gl a c j a

combination of k and yod >

O N O N O N | | | gl a k+j a [tÉs]

3. W in initial position (28) origin of Gallo-Romance w a. lat w servire > servir b. consonification malua > mauve c. Germanic *wërra > guerre additional cases, but which are different in kind d. Gallo-Romance w La Chaussée (1974: 51, 151f) in V__V: b,p > β with an intermediate stage w before o,u e. lat. -qu- [kw] (29) spontaneous sound shift w > v consensus: there was an evolutionary stage where the language lacked w altogether: a. lat w and w from consonification were already v b. Germanic w had not arrived yet La Chaussée (1974: 146) (30) Latin initial w: two paradigms lat. w > v valere valoir ventre ventre vinu vin voce voix vacca vache valle val vanu vain veela voile veena veine vendere vendre venire venir ventu vent voleere vouloir

lat. w > gw (>g) vadu gué vagina gaine vastare gâter Vasconia Gascogne *veractu guéret vespa guêpe vipera guivre viscu gui

- 12 (31) explanations offered in the literature ==> all extra-grammatical a. lexical concurrence Bourciez (1967: §163-R1), Fouché (1966-73: 559-R2-3). b. analogy Bourciez (1967: §163-R1) c. double borrowing: Gallo-Romance > Germanic > Gallo-Romance Fouché (1966-73: 559-R4) d. Celtic substrate for lat w > gw Alinei (2000:934f) (32) Germanic initial w: strengthening to gw without exception *wardôn garder *warnjan garnir *wërra guerre *wîsa guise *want gant *wahtôn guetter *waigaro guère *Wilihelm Guillaume etc.

4. W in post-consonantal position (33) lat w and w from consonification > v a. lat w w from consonification servire servir ianuariu janvier cervisa cervoise tenue ofr. tenve advenire avenir vidua veuve servier malua mauve cervariu calva chauve *cerwiice ofr. cerviz b. lat. w and from consonification is lost 1. as medial C of a triconsonantal cluster C[C]C 2. in certain cases after CC clusters and before a long vowel: CC__VV after lat. l pinguis after consonant clusters *pulw(e)re ofr. poldre (> poudre) battualia battalya *solw(e)re ofr. soldre (> absoudre) *ingwiine ofr. eine *volw(e)re ofr. voldre *victwaalia ofr. vitaille *solw(i)ta ofr. solte *fluctwaare ofr. floter *volw(i)ta ofr. volte *paskwaaticu ofr. pascage *kwerkwedula sarcelle

(> aine) victuaille flotter pacage

- 13 (34) Germanic w almost no evidence a. germ Cw > Cv one serious etymology *sparwâri > ofr. esparvier (> épervier) Fouché (1966-73: 485,696) b. germ Cw > gw one serious etymology *skârwahta > ofr. eschargaite (> échauguette) Fouché (1966-73: 699) c. a number of unexploitable candidate cases 1. écarver < old nord. skarv- ? Fouché (1966-73: 700) 2. *farwida > ofr. farde (> fard) (Fouché 1966-73: 149) *farwidan > *farwidare (> farder) (Fouché 1966-73: 495) regular loss of post-tonic i, hence of w as the medial consonant of C[C]C 3. initial sw- > sufrom vha swehur, swigar, Fouché (1966-73: 716) derives *swegaro > ofr. suegre *sweɣuru > ofr. suevre 4. Tw- > tu: þwahlja > touaille Fouché (1966-73: 693, 930) 5. fr. gaillard a host of vague and unreliable etymologies, one of which is based on got. walwjan > *gwalardo > gaillard Fouché (1966-73: 562f) 6. (loup-)garou < *werwulf ? obscure: Fouché (1966-73: 560)

- 14 -

5. Conclusion (35) summary distributional situation + = strengthening observed – = no strengthening observed yod lat. germ. #__ + + C__ +/– +

w #__ C__

lat. +/– –

germ. + (+)

(36) yod and w in Strong Position: a simple generalization a. all glides present in Strong Position at the times when strengthening by affrication became an active phonological process in the language undergo this process: yod > tÉʃ,dÉʒ w > gw b. the strengthening process is actually not affrication: strengthening of yod and w share the fact of producing a contour structure affricates gw mono-positional muta cum liquida O O O O t ʃ c.

d.

e.

d ʒ

g w

d r

contour-segment strengthening occurred "late" (i.e. in the 3rd, 4th centuries) concordant evidence 1. from yod: strengthening is "late" 2. from w: Germanic w strengthens systematically glides that escape contour-segment strengthening 1. either were not glides anymore at the relevant point in time yods that undergo metathesis, palatalization, w > v 2. or were not in post-coda position anymore, i.e. with an independent consonant to their left ==> geminate yod in modiolu > ofr moiel of course does not strengthen: it is already strong. asymmetries between yod and w are due to independent factors 1. w > v 2. w in C__: absence of exploitable evidence, but we know that contour-segment strengthening is possible: *skârwahta > échauguette 3. various things happened to yod in post-coda position because of the presence of the preceding coda ==> metathesis, palatalization, gemination there is no preceding consonant in word-initial position, hence word-initial yods are never altered and enter the strengthening period as such.

- 15 References Alinei, Mario 2000. Origini delle lingue d'Europa. Vol.2 Continuità dal Mesolitico all'età del Ferro nelle principali aree etnolinguistiche. Bologna: Mulino. Bourciez, Edouard & Jean Bourciez 1967. Phonétique française. 9e édition Paris: Klincksieck. Carton, Fernand 1974. Introduction à la phonétique du français. Paris: Bordas. siteGGHF. Fouché, Pierre 1966-1973. Phonétique historique du français. Trois vols. Paris: Klincksieck. Jacobs, Haike 1993. La palatalisation gallo-romane et la représentation des traits distinctifs. Architecture des représentations phonologiques, edited by Bernard Laks & Annie Rialland, 147-171. Paris: CNRS Editions. La Chaussée, François de 1974. Initiation à la phonétique historique de l'ancien français. Paris: Klincksieck. Pope, Mildred 1952. From Latin to Modern French with especial Consideration of AngloNorman. Manchester: Manchester University Press. Rey, Alain (ed.) 1998. Dictionnaire historique de la langue française. Paris: Robert. Ségéral, Philippe & Tobias Scheer 2001. Les séquences consonne + yod en gallo-roman. Recherches Linguistiques de Vincennes 30: 87-120. WEB. Zink, Gaston 1986. Phonétique historique du français. Paris: PUF.