Iberian Spirantization and Continuant Spreading

Ibe ian Spi an iza ion and Con inuan Sp eading Joan Masca 6 Uni e si a Au bnoma de Ba celona The spi an iza ion o oiced obs uen s in languages o he Ibe ian Peninsula is examineú, especially o Ca alan and Spanish. Phonologicai analyses based on he sp eading o he ea u e [con inuan ] ace se ious p oblems and can be man ained only a he cos o empi icai inadequacy o pos ula ion o ad hoc in e p e a ion o p inciples o addi ional ules. I is p oposed ha hese inadequacies s em om he phonologicai cha ac e o he anaiyses, and ha a phone ic ea men o he ai ema ion gi es mo e sa is ac o y esul s . The al e na ion be ween he se ies o oiced s ops [b], [d], [g], and hei espec i e spi an app oximan s [PI, [a], [y] shows a e y simila dis ibu ion in he Spanish, Ca alan, Basque, and Po uguese dialec s o he Ibe ian Peninsula. Some a ie ies o hese languages show a di e en dis ibu ion, ha may ha e o be analyzed in a di e en way; he e I will use he em "Ibe ian spi an iza ion" o e e o he dis ibu ion o s ops and spi an s desc ibed below, which is sha ed, acco ding o he a ailable li e a u e and o my own obse a ions, by mos Ibe ian a ie ies.l The de e mina ion o he p ope na u e o his dis ibu ion is in e es ing no on desc ip i e g ounds alone, bu also because o he con o e ed s a us o he ea u e [con inuan ] in phonological heo y 2, in pa icula wi h espec o i s ep esen a ional p ope ies which depend on i s possibili y o sp eading. The aim o his pape is somewha limi ed: he main goa1 is o de e mine wha he exac na u e o he dis ibu ion is, i.e. o es ablish he co ec linguis ic gene aliza ion ha co e s he dis ibu ion. Since as a as I can see mos cu en desc ip ions ely on alse assump ions (o Ca alan Wo king Pape s in Linguis ics (CWPL), 1991: 167- 179. Uni e si a Au bnoma de Ba celona else on e y simpli ied e idence), a e iew o he basic ac s migh p o e qui e use ul. Mo e speci icaly, I will each wo conclusions: a he desc ip i e le el, ha he dis ibu ion is go e ned by phone ic ac o s, and w. . . he heo y o phonology, ha , consequen ly, Ibe ian spi an iza ion cases do no cons i u e a gumen s o a sp eading ea men o [con inuan ]. 1. The basic ac s Spanish can illus a e he basic ac s. Mos desc ip ions o s anda d Cas ilian Spanish ( ollowing Na a o Tomás (1971)) p esen he ollowing s a e o a ai s (which will be e ised la e ): Pos pausall y b j6n gana des' d6n 'well' 'wins' 'dis-dain' A e nasals Ambos 6nda U'S gA 0 'bo h' 'wa e' 'a ca ' A e la e als: o d ald6a mA1 dla 'hamle ' 'bad day' b. [Pl. [al, [yl a e ocoids 6j plno blwaa $xo ( owels, glides) ' he e's wine' 'widow' 'I do' a e ica i es dezplo dezae a y Ano 'de ou ' 'si nce' 'A ghan' a e ka p6n be he m8 y w6sa 'coa1 ' 'g een' 'hea y sea' A e la e als: o b, g m l p 68 es 'a housand imes' Alga 'seaweed' As can be seen om he examples in (1) no all possible con ex s a e a es ed in Cas ilian Spanish; in pa icula , he e a e no examples o b, d, g, a e obs uen s s ops. This is so because no obs uen s ops appea syllable inally; in his posi ion hey a e no mally weakened and spi an ized ( 68 pol, e c.). These con ex s a e ound in Ca alan, howe e . This language shows he same dis ibu ion o [b], [d], [g] s. [PI, [a], [y] as in (1) (see, e.g., Wheele (1979)). Bu , as opposed o Spanish, obs uen s ops a e ound syllable inally, and in his con ex he noncon inuan e sions [b], [d], [g] show up; no e ha he e is also eg essi e oicing assimila ion, hence no [pb], [kd], e c.: (2) ubb51 'socce ' sClbdi 'subjec ', ' assal' k 6b g% 'no ca ' s69 b5 'I am good' ma egda 'eme ald' p5 g gds 'li le as e 2. The s anda d gene aliza ion Thus s anda d desc ip ions ha e lead some au ho s (Goldsmi h (1981), Masca ó (1983), Hanis (1984,1985), Hualde (1988), Rome a (1990), Palmada (199 1)) o p opose ha spi an iza ion can be cap u ed by he gene aliza ion in (3). (3) Spi an P, 8, 8 appea a e [+con inuan ] segmen s, s ops appea a e [-con inuan ] segmen s and a e pause. When leshed ou in some pa icula heo y, (3) akes he o n o se e a1 a ian s, depending on whe he only P, 8, ).j, (Hanis (1984)) only b, d, 9, (Hualde (1988)) o bo h (Lozano (1979), Masca 6 (1983), Hams (1985), Palmada (1991)) a e de i ed by ule. In he amewo k o au osegmen al phonology (3) has de e mined, as can be expec ed, he p oposa1 o a sp eading ule ha associa es delinked o unspeci ied b, d, g o he [con ] alue o he p eceding segmen ; in hose cases in whch no sp eading akes place, he unma ked [con ] alue is assigned no mally by de aul . The basic p oblem aced by analyses based on he gene aliza ion in (3) as applying o he da a in (1) is he odd beha io o la e als. The la e al de e mines he appea ance o spi an s in he case o b, g, bu o a s op in he case o d. (m p60esJ Alga s. ald6a). Bu [I] should be ei he [+can ] o [-con ], and we would expec consequen ly a uni o m solu ion, i.e., ei he ld, * lb, * 19 o * 18, lp, 18, which is w ong in ei he case. Le 's call his he "la e al p oblem." Al hough I don' wan o en e in o he phonological ea men s in de ail, I hink ha i is ai o say ha , as shown by he di e si y o ea men s a ailable, he e a e no sa is ac o y solu ions. This is e en mo e he case i he ac s a e examined in mo e de ail (c . below). 3. The issue o homo ganici y. Almos all phonological analyses, since Hanis (1%9), ha e ied o sol e he la e al p oblem by de i ing he asymme y om he homo ganic cha ac e o he clus e ld. This has been done by including a condi ion o sameness o place in he ule i sel , o by de i ing he e ec s om mo e gene al p inciples.3 In his sec ion I wan o show ha a "homo ganic solu ion" is w ong on a ac ual basis. No ice i s ha in Spanish, as shown in (I), he co ela ion be ween lack o spi an iza ion and homo ganici y holds owing o a lack o e idence: he e a e cases o ho no ganic clus e s o nasal o la e al + s op (m b, nd, 99, ld) which obey he gene aliza ion, bu no a es ed cases o he e o ganic clus e s o nasal o la e al + spi an . In o he wo ds, he homo ganic solu ion makes a speci ic he p edic ion, namely ha oiced s ops a e he e o ganic nasal o la e al will spi an ize, bu he p edic ion canno be checked because nasals and la e als only appea assimila ed in place o he ollowing oiced s op. Mo eo e , his lack o coun e examples o he homo ganic solu ion is mo e appa en han eal. In he case o clus e s wi h an ini ial ica i e we can ge €Ja (p og essi e assimila ion o place and eg essi e assimila ion o oice): em be@e 'ins ead o ', and al hough i is clea ly homo ganic wi h he p eceding consonan , he d ge s spi an ized. Bu e en in he case o nasals and la e als, i we go beyond he no mal cases, as hose epo ed in (I), he e is coun e e idence o he "homo ganic solu ion". I a p onuncia ion wi h he e o ganic clus e s is o ced on speake s, we ne e ge he spi an --as a "homo ganic solu ion" would p edic . This canno be done wi h all he cases, bu is qui e na u al o clus e s like md, mg, gb, gd, in which case he s op appea s exac ly as in he cases whe e homo ganici y holds. I should be poin ed ou ha in all "homo ganic" analyses homo ganici y is c ucial o la e als, bu only in some (e.g. Hualde (1988)) o nasals as well. (4) En Vie na m d]eí Su 'in Sou h Vie nam' E1 Isla[m glue e o ' he igh ing Islam' pa lu [ ;] b] a a o 'cheap pa king' go[g dle pla a 'sil e gong' In he case o Ca alan i is much easie o show ha spi an iza ion does no depend on homo ganici y: he e o ganic clus e s o o al o nasal s op + b,d,g a e qui e equen , and include also la e als. (5) a. ~Clbdi b. s6m d6s 'we a e wo' c. g BA d ndi ' u key' keb ga s6m g 6ns 'we a e big' b6Adu 6 J (place name) s69 b5 6p b5 'good yea ' up6A d5w 'egg yoik' ma agda BJ db ' ough yea ' uAdapSw 'po hole' ep g an 'bis yea ' s y~ d as ' i e days' b ~ d S a 'I come ea ly' g dhlpa 'sun ise' 4. Back o he ac s. The ine icacy o he homo ganic solu ion lea es us wi h he la e al p oblem unsol ed. In addi ion, analyses based on he gene aliza ion in (3) ace u he p oblems when he empi ical domain is checked somewha mo e ca e ully. The e a e mainly wo p oblems o be added: i s , he "di e ences in a iabili y" ha de e mine an asymme y be ween con ex s showing ixed solu ions, and con ex s ha allow bo h spi an s and s ops in a ia ion; second, he addi ional odd case o ica i e+b,d,g clus e s, whe e we ge a dis ibu ion simila o ha o la e als. Some o he li e a u e igh ly s esses he a iabili y o he phenomenon: in some cases he solu ion p esen ed in (1) is dominan , bu no unique, and spi an and s op can bo h appea in some con ex s.4 Second, con ex s do no beha e equally wi h espec o a iabili y. In some cases he e is ha dly any a ia ion: he o ms in (6a) a e o ms ha a e no only e y a e s a is ically in on o he spi an ized o ms in (6a'): when p esen ed wi h such sequences, speake s ha e s ong nega i e judgemen s o accep abili y. On he o he hand, o (6b) hose judgemen s ange om dubious o accep able, when compa ed wi h he mo e no mal cases o (6b'). (6) a. * aba *ada * aga a.' apa aaa aya *a npa * ancla * agya * alaa amba anda agga alda b. (?)axba (?)asda (?)sega b.' axpa asaa aeya5 (?>Pa (?)3a (?)ga ba da ga In he absence o mo e de ailed expe imen al da a, he a iabili y can be summa ized as ollows. I indica e he di e en con ex s (7a-k) and he con inuan alues o he elemen o he le o b, d, o g, and o b, d, g hemsel es, and I ha e added an example. CLUSTER OBSTRUENT STOP NASAL VOCOIDS, R LATERAL + D LATERAL + B , G PAUSE FRICATIVES + B D G: F+ B CJlTBR, HOMORGANiC F + D,G Chl-EF HEXEBORGAMC CONT VALUE [con ] 1s [con ] 2nd - - EXAMPLE sClbdi (Ca .) 6nda bIw Sa ald6a alga gana bC b Gs k (Ca .) dezae bCl 3i 6k a (Ca .) dezplo (7) inco po a es he obse a ion, o mal1 y epo ed a leas o Ca alan (Recasens ( 1986)), ha in he case o he clus e b we no mally ind he s op ([ b]). Acco ding o my own obse a ions wi h se e al speake s, he s op is much mo e a o ed in he case o @ han in he o he cases o ica i e + b, d, g., i.e.: (8) a. b>p b. 8>d c. -9 bu b usc bu di ec e bu glacial ilbso basc ilbso disc e ilbso gallec iom e bal iom doble iom glo ids The da a summa ized in. (7) show he ollowing p oblems o an analysis based on sp eading o [con inuan ]. Fi s he la e al p oblem, as al eady no ed. Second, he ac ha , al hough homo ganici y has been shown no o play a ole in he case o nasals, la e als and s ops, in he case o ica i es a close place o a icula ion makes he s op, i no s ic ly necessa y, a leas p e e ed (7h), o possible (7i). Thi d, why should he e be a iabili y in pos pausal posi ion, and a e homo ganic o quasi-homo ganic ica i es? I is a he s ihng ha in he cases in which he [con inuan ] alue is assigned by de aul , as o owels and nasals, o unclea (la e als) we ge a ixed solu ion, whe eas in hose cases whe e i is unde lying ( ica i es) we ge a iabili y. Dialec s showing Ibe ian spi an iza ion di e widely in su ace esul s depending on he in e ac ion o o he p ocesses, bu all show he kind o ixed beha io o (3) o (7). Le 's examine he la e al p oblem i s . And le us suppose o a momen ha we abandon he equi emen o a s ic phonological solu ion. Is he e any hing in common, e.g. be ween clea s ops, as in (7a, b), and [I] o [A] ollowed by [d]? Is he e any hing in common, simila ly, be ween clea con inuan s, like owels o glides (7c), and [I] o [8] ollowed by [PI o [a]? A ca e ul examina ion o he execu ion o a sequence like [ld] shows ha he a icula o y co ela es o [+can ], ai low h ough an open ocal ac hold ue o all he ocal ac , excep o i s cen al pa , in he pala oal eola egion. In his a ea he ai low is in e up ed by he con ac o he on pa o he ongue blade and he ee h. Lowe ing o he back o he ongue blade allows he ai o escape a ound he cen al closu e, be ween he cheek and he ee h. A phonological solu ion imposes /l/ o be ei he [+con inuan ] o [-con inuan ], bu phone ically he co ela es o his ea u e a e no equally dis ibu ed along he ocal ac . Fo /l/ he e is ai low om he glo is o he lips, bu he den o-pala o-al eola egion is excluded. The same holds ue o / i, which di e s om he o me because he lingual con ac is ex ended u he back. In a ce ain sense we a e ex ending he idea ha nasals, hough clea ly con inuan s in a global sense, a e o al s ops, o la e als, ha --N.B., phone ically-- hough con inuan s in a global sense, a e s ops in he cen al al eopala al egion. The gene al idea is hus ha a) no phona ion implies (unde no mal ci cums ances) NO ai low; b) o al o al closu e implies NOai low (o close o ze o i he e is a p essu e inc ease); c) o al o al ape u e (du ing phona ion) implies ai Jow.; d) local closu es de e mine ai low in gene al, bu NO ai low locally.. The e is an ob ious way o cap u e he ideas ske ched abo e. Spi an iza ion is co ela ed wi h exis ence o ai low du ing he p oduc ion o he p eceding segmen . Bu , c ucially, his ai low has o be egis e ed a he place o a icula ion o he spi an izing consonan , no a he place o he p eceding segmen i sel . In o he wo ds, [d] is a s op because a i s place o a icula ion, i.e. he den al egion, he e is no ai low du ing he a icula ion o [I]. On he o he hand, [PI and [VI a e spi an s because a i s place o a icula ion, i.e. he labial and he ela egions, espec i ely, he e is ai low du ing he a icula ion o [I] The e is an ob ious way o cap u e he ideas ske ched abo e. Spi an iza ion is co ela ed wi h exis ence o ai low du ing he p oduc ion o he p eceding segmen . Bu , c ucially, his ai low