690 likes | 839 Views
41st Linguistic Symposium on Romance Languages Ottawa, May 5-7, 2011. Branching Onsets in the light of Sardinian metathesis and diachronic lenition in French. Tobias Scheer Université de Nice – Sophia Antipolis, CNRS 6039. Gvt. in strong position C__V. in weak position V__V. Gvt. Lic.
E N D
41st Linguistic Symposium on Romance Languages Ottawa, May 5-7, 2011 Branching Onsets in the light of Sardinian metathesis and diachronic lenition in French Tobias Scheer Université de Nice – Sophia Antipolis, CNRS 6039
Gvt in strong position C__V in weak position V__V Gvt Lic Lic <== IG <== IG purpose of the talk: to convince you that this is the identity of muta cum liquida (a branching onset)
roadmap introduction to the theory: the Coda Mirror and (syntactic) locality (Relativized Minimality) • three phenomena are examined • lenition of muta cum liquida • (Celtic, Latin > Italian, Gorgia Toscana) • Latin > French • Gallo-Romance: ALF (French, Occitan, Franco-Provençal) • compensatory lengthening within (sic) a muta cum liquida: • lat. latroone > fr. laRRon • metathesis in Tertenia Sardinian: • /dormendu/ dromendu
The Strong Position in Phonology • the Strong Position • (in Romance and elsewhere, Ségéral & Scheer 2001, 2008) • - {#,C}__ = Strong Position: PORTA > porte • TALPA > taupe • - V__V = weak position A: FABA > fève • - __{#,C} = weak position B (Coda): LUP(U) > l[u] • RUPTA > route • the mirror effect: {#,C}__ vs. __{#,C} are symmetric • - with respect to their position: mirror image • - with respect to their effect: strength vs. weakness
Gvt Gvt Lic Lic • relevant consonants: • the word-initial consonant • the consonant that occurs after a coda the initial CV = # represents the morphological information « beginning of the word » The Coda Mirror: Government & Licensing • analysis in CVCV (Lowenstamm 1996, Scheer 2004) # P O RTA T A L P A • consonants in Strong Position are • licensed • but ungoverned consonants in Strong Position occur after an empty nucleus ø __ • Government • inhibits the segmental expression of its target • empty nuclei must be governed • Licensing • promotes the segmental expression of its target
Gvt Gvt Gvt Lic Lic Lic The Coda Mirror: Government & Licensing • intervocalic V__V: the consonant is • not adjacent to any empty nucleus • licencensed and governed • in coda position: the consonant • occurs before an empty nucleus: __ø • is neither licensed nor governed F A B A R U P T A L U P (U)
The Coda Mirror: Government & Licensing • summary • Strong Position = {#,C}__ = ø__ = strength = ungoverned but licensed • Coda = __{#,C} = __ ø = weak A = ungoverned and unlicensed • intervocalic = V__V = V__V = weak B = governed and licensed
Gvt branching onset Lic <== IG branching onsets in CVCV • worse than making a wrong prediction: making NO prediction at all • what a branching Onset looks like (after a consonant) • the solidarity of the cluster is due to a relationship that is contracted by the (melodies of) the two consonants: IG (Infrasegmental Government) • the liquid R: licensed, but ungoverned ==> strong position • the obstruent T: target of neither Gvt nor Lic ==> ??
Locality in Syntax Relativized Minimality, Rizzi (1990) given two classes of items A and B, a relation between A1 et A2 is local iff no other A intervenes
Couldi John __i have come ? Havei John could __i come ? head head head head head head argument argument * Locality in Syntax • three major classes of items in syntax: • - verbs (heads) • - arguments (A position) • - quantifiers (A’ position) ☺ John could have come John could have come
branching onset Lic A M P L U S <== IG Locality in Syntax • a branching onset is a non-local structure: • - major classes of items in phonology are: onset and nucleus - an internuclear relation exists whereby a third nucleus intervenes. Gvt
The trouble: summary • no prediction made • 2. violation of locality
Gvt branching onset Gvt Lic <== IG The cure: making branching onsets local instead of having a non-local government relation • the intervening nucleus is the source of government • it is entitled to govern because it is not itself governed: it is unpronounced for a different reason (IG) • consequence: the definition of what a good governor is owes nothing to phonetics • before: only nuclei with phonetic content can govern • now: a nucleus can govern iff it is not governed itself
Gvt TR in Strong Position TR in intervocalic positon Gvt Lic Lic <== <== local branching onsets: predictions • when the TR is preceded by an empty nucleus (Strong Position), the T will also be in Strong Position (licensed but ungoverned) • in case the TR is in intervocalic position, the T will also be in intervocalic position (licensed and governed)
local branching onsets: predictions • hence the following prediction: • the T of a TR group behaves exactly like a simplex T • - if the TR group is in Strong Position, T will be strong • if the TR group is in intervocalic position, T will be intervocalic in other words: given a branching onset TR, T behaves like if R were not there
testing the prediction • typologically speaking, branching onsets are rare • even rarer are languages that allow to test the reaction of TRs on lenition • ==> the empirical situation largely understudied • we examine 5 cases: • - voicing in Latin > Northern Italian dialects • - Celtic (in its prehistory) • - Gorgia Toscana • - French diachrony • - Gallo-Romance dialects as witnessed by the ALF • (Atlas Linguistique de la France)
testing the prediction in each case, T behaves alike in V.TRV = V.TV C.TRV = C.TV
Grande GrammaireHistorique du Français(GGHF) env. 1800 pages projet en cours livraison prévue vers 2013-14 éd. Christiane Marchello-Nizia Bernard Combettes Sophie Prévost Tobias Scheer
test case 1: Northern Italian dialects (lomb., lig., venez. etc.) e.g. Rohlfs (1966:§260) "I gruppi cr, tr, pr. In questi gruppi la consonante occlusiva viene trattata esattemente come se si trovasse in posizione intervocalica."
test case 2: Celtic the classical scenario assumes 3 stages (e.g. McCone 1996) • stage 1: IE b,d,g > v,,ɣ / V__V et V__RV • V__V • IE Proto-Celtic Old Irish glose • b kladibos *klaivos klaiəv épée • d kladibos *klaivos klaiəv épée • g tegos *teɣos tieɣ maison • 2. V__RV • b dubro- *duvro- dovər eau • d widwa: *wiwa: fiev veuve • g wegros *weɣros fe:r herbe • 3. but resistance in Strong Position {#,C}__ and in gemination • N__ *windos fiind blanc • #__, gém *buggos bog mou
test case 2: Celtic • stage 2: as stage 1, but now also across word boundaries • stage 3: t,k > , / V__V and V__RV (there is no p) • V__V • Insular Celtic Proto-Irish Old Irish glose • t *ehja teɣah *eja eɣa ə ieɣ sa maison • k *inda: kloka: *inda: loa: iŋ lo la pierre • 2. V__RV • t *bre:tra: *bre:rə briiaər mot • k *dakra *dærə die:r larme • 3. but resistance in Strong Position {#,C}__ and in gemination • R__ *eisko- *eisk iask poisson • gém *makwkwos *makwkwah mak garçon
test case 3: Gorgia Toscana Castellani (1960), Giannelli & Savoia (1978, 1979), Marotta (2000-01, 2008) p,b,t,d,k,g > ɸ,β,θ,,x/h/ø,ɣ / V__(R)V • V__V • Stand. It. Tuscan glose • p apɛrto aɸɛrto ouvert • t laato laaθo côté • k bruuko bruuxo, bruuho, bruuo worm • 2. V__RV • p la piega la ɸjɛɛɣa le pli • t liitro liiθro litre • k la krɛɛma la xɾɛɛma, la hrɛɛma la crème • 3. but resistance in Strong Position {#,C}__ and in gemination • R__ pɔrta pɔrta porte • #__ pjɛɛde pjɛɛe pied • gém. gatto gatto chat
v v p b p b p p v v b b test case 4: French • only labials and dentals are examined – the situation of velars is complicated by palatalizations (Bourciez 1967 etc.) • labials in TR groups • simplex Labials
t t ø ø d d t t ø ø d d test case 4: French • dentals in TR groups • simplex dentals
test case 5: Gallo-Romance dialects (ALF) this part of the talk is joint work with Guylaine Brun-Trigaud ALF Atlas Linguistique de la France Gilléron, Jules, and Édmond Édmont 1902-10. Atlas linguistique de la France. Paris: Champion, 9 vol., supplément 1920. based on fieldwork 1897-1900, 639 points of inquiry.
test case 5: Gallo-Romance dialects (ALF) • prediction in a dialectal system • T alone and T in a TR cluster behave alike in every given system (dialect) • hence • for each obstruent and each position, the isoglosses of T alone and T in a TR cluster are identical. ==> not exactly a trivial or intuitive prediction ==> a prediction about 639 systems at the same time • examination of labials in intervocalic position • dentals are inconclusive for independent reasons (desolidarisation, see next slide), velars are blurred by palatalisations.
test case 5: Gallo-Romance dialects (ALF) • goal: comparison of • -P- with -PR- • -B- with -BR- • variation and its interpretation: • only actual branching onsets (solidary TR groups) are an input for the comparison. Hence non-solidary groups are counted out: • coda vocalisation betrays desolidarisation: V.TRV > VT.RV • (grey-shaded on the maps below) • example: • solidary TR group: FEBREM > fièvre, TAB(U)LA > table • non-solidary TR group: FEBREM > fewre, TAB(U)LA > tole
level 0 level 1 level 2 B 0 B 1 B 2 ? BR 0 BR 1 B’R 0 BR 2 BL 0 BL 1 B’L 0 test case 5: Gallo-Romance dialects (ALF) -B- vs. -BR- syntheses lexical basis ALF -B- ABANTIARE > avancer ABOCULUS > aveugle FABA > fève HIBERNU > hiver -BR- (primary) FEBREM > fièvre -B’R- (secondary) BIB(E)RE > boire SCRIB(E)RE > écrire -BL- (primary) OBLITARE > oublier -B’L- (secondary) SAB(U)LU > sable DIAB(U)LU > diable STAB(U)LA > étable
? test case 5: Gallo-Romance dialects (ALF) -P- vs. -PR- syntheses lexical basis ALF level 0 level 1 level 2 -P- CREPANT > crèvent NEPOTE > neveu *ARRIPARE > arriver TROPARE > trouver LUPA > louve SAPONE > savon SAPA > sève -PR- (primary) APRILE > avril -P’R- (secondary) PIP(E)R > poivre LEP(O)RE > lièvre OP(E)RARIU > ouvrier -PL- (primary) DUPLU > double -P’L- (secondary) CAP(U)LU > câble P 0 P 1 P 2 PR 0 PR 1 P’R 0 PR 2 PL 0 PL 1 P’L 0
-P- alone intervocalic
-P- in an intervocalic TR group
poitevin Croissant superposition: intervocalic -P- alone and in a group
test case 5: Gallo-Romance dialects (ALF) • Croissant • -P- alone spirantizes (-P- > -v-), but remains a stop in • -PR- (> -br-). • 8 points of inquiry (503‑5, 600, 601, 800, 802, 803) • well-known zone of transition (Croissant).
test case 5: Gallo-Romance dialects (ALF) • poitevin • opposite pattern: P in PR spirantizes (> vr), but lexical variation is observed for isolated P (> b, v) (which however always voices) • 24 points of inquiry (429, 448, 459, 479, 509-13, 515, 517, 518, 521, 525, 527-29, 533, 535, 536, 540, 621, 630, 632) • since ‑PR‑ always spirantizes, a fricative output is also expected for ‑P‑. • lexical basis ALF: 7 words • unexpected non-spirantization concerns only two words • two contravening words: LOPA > loube, SAPONE > sabon • LOPA: contravening in 19 out of 24 points, SAPONE in 17 out of 24. The five other words are well-behaved in all 24 points. ==> lexical inconsistency points to contact, rather than to regular evolution.
-B- alone intervocalic
-B- in an intervocalic TR group
provençal superposition: intervocalic -B- alone and in a group
test case 5: Gallo-Romance dialects (ALF) provençal - there are no *vl, *vr at all - hence -BL-, -BR- could not produce *vr, *vl (cf. *vl in oïl)
compensatory lengthening TR > RR in French, lat. tr,dr (primary and secondary) lose their dental with eventually ensuing compensatory lengthening of the r. Fouché (1966-73:719ff) etc.
compensatory lengthening TR > RR • three factors • monophthongs vs. diphthongs • light (ie, ue) vs. heavy (ei, ai) diphthongs • stress for monophthong
compensatory lengthening TR > RR Latin vowel length is irrelevant
compensatory lengthening TR > RR only Gallo-Romance vowel length matters G-R length is stress: tonic vowels (in open syllables) are long unstressed vowels are short
compensatory lengthening TR > RR • (Gallo-Romance) stress does not matter for diphtongs • heavy diphthongs always prohibit gemination • light diphthongs always provoke gemination (but are also always tonic) • ==> heavy diphthongs are inherently long • ==> light diphthongs are inherently short
1) tr,dr > r / VV__ compensatory lengthening TR > RR 1. tr,dr > r / VV__
1) tr,dr > r / VV__ compensatory lengthening TR > RR 2. tr,dr > rr / V__
1) tr,dr > r / VV__ compensatory lengthening TR > RR 2. tr,dr > rr / V__ (choice of relevant data for < V unstressed)
1) tr,dr > r / VV__ compensatory lengthening TR > RR alternations within verbal paradigms
1) tr,dr > r / VV__ compensatory lengthening TR > RR generalization tr,dr > r / VV__ tr,dr > rr / V__ ==> *VVC.CV superheavy rhymes are forbidden ==> this much we knew anyway
1) tr,dr > r / VV__ compensatory lengthening TR > RR • *VVC.CV • this much we knew anyway: • this is why there is a contrast between the evolution of vowels in open and closed syllables • stress is converted into length, but long vowels can only exist in open syllables • ==> Vstr.CV VV.CV • vs. VstrC.CV VC.CV
1) tr,dr > r / VV__ compensatory lengthening TR > RR • compensatory lengthening • how can • an onset-x slot become a coda ? • worse: a non-moraic consonant become moraic?