Full text
Abstract The aim of this paper is to account for the similarities and differences between Eastern Catalan and Valencian truncated hypochoristic words in the framework of Correspondence Theory (McCarthy & Prince 1995), specifically Transderivational Relation Theory or output-to-output correspondence relation (Benua 1995, 1997). The prosodic constituent determined by lexical stress in Catalan makes up the most of the hypochoristic truncated forms in both dialects. Faithfulness to the base in Eastern Catalan can make longer forms than the left edge of the metrical constituent imposes; instead unmarked conditions on the shape of Valencian truncated forms can force the insertion of the gender vowel, thus increasing the metrical constituent at its right edge. In addition, some Catalan hypochoristics furnish empirical evidence for the full model of Correspondence Theory since they require underlying information of the base form. Key words: dialectology, Catalan, truncation, correspondence theory. Resum. Fidelitat als límits prosòdics. Variació dialectal en els mots truncats del català L’objectiu d’aquest article és donar compte de les similituds i diferències entre els mots hipocorístics truncats del català oriental i els del valencià en el marc de la Teoria de la correspondència (McCarthy i Prince 1995), específicament de la Teoria de la relació transderivacional o relació de correspondència entre outputs (Benua 1995, 1997). El constituent prosòdic determinat per l’accent lèxic en català constiueix la majoria de les formes hipocorístiques truncades en tots dos dialectes. La fidelitat a la base en català oriental pot formar formes més llargues que les que imposa el límit esquerre del constituent mètric; en canvi les condicions no marcades sobre la configuració de les formes truncades valencianes poden forçar la inserció de la vocal de gènere i així allarguen el constituent mètric en el seu marge dret. A més, alguns hipocorístics catalans aporten * A first version of this paper was presented at the 1998 Colloquium on Generative Grammar at Palmela (Portugal). This work has been supported in part by research grants from the Ministerio de Educación y Ciencia DGES PB96 1199-CO4-04 and the Generalitat de Catalunya 1997 DGR 00033 and 1998 XT-00065. I wish to thank the anonymous reviewers as well as Nicole Nelson, Teresa Vallverdú, Maria-Rosa Lloret, Joan Mascaró, Eulàlia Bonet, Jaume Mateu, Jaume Solà, Xavier Villalba, M. Teresa Ynglès and Michael Kenstowicz for comments and criticism. CatWPL 6, 1998 7-22 Faithfulness to Prosodic Edges. Dialectal Variation in Truncated Words in Catalan* Teresa Cabré Universitat Autònoma de Barcelona. Departament de Filologia Catalana 08193 Bellaterra (Barcelona). Spain [email protected].es Received: December 13th 1998 Accepted: March 17th 1999 CatWPL 6 007-022 21/10/99 12:44 Página 7
evidència empírica per al model ple de la Teoria de la correspondència ja que requereixen informació subjacent de la forma base. Paraules clau:dialectologia, català, truncació, teoria de la correspondència. 1. Preliminary considerations As is well known, all dialects of Catalan delete the underlying final /n/ of nouns and adjectives in which stress falls on the final syllable. This phenomenon can be explained in Correspondence Theory by a specific condition of No Coda that dominates the input-to-output faithfulness conditions. This underlying final /n/ constitues the second mora of the final heavy syllable. Except for some marked cases, Catalan needs a final heavy syllable to have final stress1. This final /n/ appears in plurals and in derived words: camí «path», but camins, caminet, caminar, caminal. We show this phenomenon in (1). (Only the syllabification and stress assignment is specified; otherwise the standard spelling is used). On the other hand, a similar deletion of final /R/ occurs in Eastern Catalan and North-Western Catalan but not in Valencian2. Final /R/ is also deleted before the plural morpheme, but appears in derived words: pape(r) «paper», pape(r)s, pape8 CatWPL 6, 1998 Teresa Cabré 1. The final stress in camí<n> and papé<r> is an opacity effect because two moras are needed to have final stress but the second mora is deleted. 2. North-Western Catalan and Valencian have been considered traditionally in the same dialectal group. Table of Contents (1) input camin No Coda (final n) Faith -IO ca.mín *! ☞ ca.mí * input Damian No Coda (final n) Faith -IO Da.mi.án *! ☞ Da.mi.á * CatWPL 6 007-022 21/10/99 12:44 Página 8 1. Preliminary considerations 2. The data 3. The model 4. Analysis 5. Cases from final syllable stressed bases 6. Valencian final syllable stressed bases 7. Final comment References
ret, empaperar, paperassa, paperer. It is shown in (2). (The standard written form of these words is with rin all dialects). This phenomenon is reflected in truncated words like Ció, Tià, Vador or Tasar. All are unmarked final syllable stressed words with the final consonant deleted. 2. The data The dominant characteristic of the truncated hypochoristics3in both dialects consists of the fact that the truncated form corresponds to the metrical foot, settled on the right edge of the base, which holds the word stress. So the stressed syllable of the base is always the stressed syllable of the truncated form. If the metrical foot is disyllabic, the truncated form mimics it in all dialects. We will try to show why Eastern Catalan has monosyllabic and disyllabic truncated hypochoristics when the bases end in a stressed heavy syllable, whereas in Valencian all truncated forms are leftheaded and disyllabic. On the other hand, the differences between the truncated forms and the correspondent part of the bases in both dialects are the result of the dominance of two wellformedness phonological requirements: the truncated words must have initial onset and do not accept complex onsets. The examples in (3) all have the same shape in both dialects: ‘LL and ‘HL. The Prosodic Word coincides with the stress foot of the base. The lack of a hypochoristic form in one of the two dialects is not relevant, since it can be due to a lack of information or a lack of use in a particular dialect. (All names are presented with their syllabic limits, but some of them can show dialectal differences.) (3) Base name Eastern Cat. Valencian Er.nes.ti.na Ti.na Ti.na A.gus.ti.na Ti.na Ti.na En.ri.que.ta Que.ta Que.ta Dialectal Variation in Truncated Words in Catalan CatWPL 6, 1998 9 3. Modern Catalan exhibits a different kind of truncation at the left edge of the base (for example, Montse from Montserrat). This type of truncation constitues a different system of the one analysed in this paper. It is not possible to compare it to the French system in which Anchor left and Anchor right compete. See N. Nelson (1998) for an analysis of French hypochoristics in terms of Correspondence Theory. (2) input paper No Coda (final r) Faith -IO pa.pér *! ☞pa.pé * input Roser No Coda (final r) Faith -IO Ro.sér *! ☞Ro.sé * CatWPL 6 007-022 21/10/99 12:44 Página 9
(3) Base name Eastern Cat. Valencian Jo.se.fi.na Fi.na Fi.na Jo.se.pa Pe.pa Pe.pa Te.o.do.ra Do.ra Do.ra Ge.no.ve.va Ve.va — Ma.no.li.ta Li.ta Li.ta Ca.ro.li.na Li.na Li.na Ce.les.ti.no Ti.no Ti.no Mar.ga.ri.ta Mi.ta Mi.ta Je.ro.ni No.ni — Do.min.go Min.go Min.go Rai.mun.da Mun.da Mun.da Fran.cis.co Cis.co Cis.co Clo.til.de Til.de — Vi.cen.ta Cen.ta Cen.ta Le.o.pol.do Pol.do Pol.do Ce.cí.lia Ci.la — Eu.là.lia La.ia — Mi.se.ri.còr.dia Co.ia — Ro.sa.rio — Sa.ri An.tò.nia To.na To.na The set of examples in (4) is typified by the fact that all the bases end with a stressed heavy syllable. The shape of the truncated outputs changes: while in Valencian all the examples are leftheaded disyllabic Prosodic Words, in Eastern Catalan the truncated forms, much more faithful to the base, can take one or two syllables from the base. Cases like Remei or Joaquim can only produce monosyllabic hipochoristics, cases like Bartomeu or Montserrat make disyllabic and monosyllabic truncates, and finally other bases like Daniel or Sebastià only make disyllabic ones. (4) Base name Eastern Cat. Valencian Re.mei Mei — Mi.quel Quel Que.lo Meritxell Txell — Vi.cent — Cen.to Llu.ís — Ui.sso I.sa.bel Bel Be.la Jo.a.quim Quim Xi.mo Se.gi.mon Mon — Mont.se.rrat Se.rrat, Rat — Bar.to.meu To.meu, Meu — Bal.ta.sar Ta.sar, Xar Sa.ro Sal.va.dor Va.dor Vo.ro Is.ma.el — Me.lo 10 CatWPL 6, 1998 Teresa Cabré CatWPL 6 007-022 21/10/99 12:44 Página 10
(4) Base name Eastern Cat. Valencian Da.ni.el Ni.el Ne.lo Con.cep.ció Ci.ó — Se.bas.ti.à Ti.à — It is interesting to note that hypochorization in Eastern Catalan is always a shortening process of the bases whereas in Valencian it can be understood as a way to make leftheaded disyllabic Prosodic Words. Thus disyllabic iambic bases hypochorize in disyllabic trochaic forms: Vicent → Cento, Manel →Nelo, Rafel →Felo, Lluís →Uisso.4 The possible shapes of truncated forms in Eastern Catalan are: ‘LL, ‘HL, ‘H and L’H; only a subset of them are possible in Valencian: ‘LL and ‘HL. 3. The model Benua (1995, 1997) adapts the basic model of Correspondence Theory (McCarthy & Prince 1995) to the truncation process. The truncated forms are only dependent on the outputs of the base words because the outputs are needed to produce truncated words. We are dealing with a transderivational or output-to-output relation. The base outputs are the inputs of the truncated words.«Truncated forms do not stand in a correspondence relation with the input» (Benua 1995:122). Benua’s schema is presented in (5): This proposal accounts for two questions raised in an input-to-output relation (Cabré, 1993). On the one hand, truncated forms cannot strengthen the initial flap of the base foot in cases like Je[ɾ]oni —> *[r]oni, or Marga[ɾ]ita —> *[r]ita, or Ma[ɾ]ia —> *[r]ia. On the other hand, Catalan cannot truncate proparoxiton bases like Angèlica or Penèlope, but forms like Laia, Coia, Sari, Tona are very common in spite of their proparoxiton base inputs Eulàlia, Misericòrdia, Rosàrio, Antònia. Only a transderivational or an output-to-output relation can explain why a trill does not fulfil the BT-Identity relation with a correspondent flap. We can account for forms like Laia or Coia, given the impossible proparoxiton forms, if the stress foot is disyllabic like the base output. Starting from the base input, repairing strategies are needed which are ruled out by Optimality Theory (Prince & Smolensky, 1993). Truncated words must be faithful to the surface prosodic and segmental shape of their bases. Onset conditions on the truncated outputs force the differences between Dialectal Variation in Truncated Words in Catalan CatWPL 6, 1998 11 4. One finds Pep and Pepo from Josep in Valencian, but monosyllabic hypocoristics must be understood as lexicalized forms since they are not generalized. (5) BT-Identity Base Truncated Form IO Faith Input CatWPL 6 007-022 21/10/99 12:44 Página 11
them. These cases constitute empirical evidence in favor of the output-to-output correspondence relation since the truncated forms appear completely dependent on the output base. Truncated forms in Catalan have been analyzed as templatic targets that copy the positive Prosodic Circumscription of the bases (Cabré, 1993, 1994). They also have been analyzed under Optimality Theory, with alignment and parsing constraints which make useless the old repairing strategies to adapt the Prosodic Circumscription limits to templatic targets (Cabré & Kenstowicz, 1995). Under Correspondence Theory, the faithfulness to the Prosodic Circumscription is changed in terms of prosodic faithfulness requirements. «Foot faithfulness is never direct, it is always mediated by segments bearing head or edge roles in the foot. Constraints of the Anchoring family (successors to Alignment) are responsible» (McCarthy, 1998:10). Following this proposal, we take the notion of Anchor-Position to characterize the faithfulness to the head and the edges of the prosodic constituent formed by the stressed foot of the base, which is settled on the right edge of the word, as is well known. Anchor-PosBT(FT, head): The prominent element of the truncated form and the prominent element of the base form stand in correspondence relation. Anchor-PosBT(FT, initial): The initial segment of the prosodic constituent of the base stands in correspondence relation with the initial segment of the truncated form. Anchor-PosBT(FT, final): The final segment of the prosodic constituent of the base stands in correspondence relation with the final segment of the truncated form. We must keep in mind that lexical stress in Catalan is assigned under a moraic trochee pattern.5The shape that determines the assignment of word stress can be modified in the output form. As the final /n/ and the final /r/, which constitute the second mora of the monosyllabic foot, are deleted, the stressed constituent has only one mora in the output form. On the other hand, no onset position is needed to fulfil the metrical requirements, but onset conditions can compel the output form to fill this position, even with two segments, as in Con.cep.ci.ón →Con.cep.ció. In this example -ció is the prosodic constituent of the base output on which the Anchor-Position constraints operate. As Benua states «in languages with contrastive stress, lexical prosodic structure is posited, but the surface prosody of the word must still be determined by the interaction of faithfulness constraints with metrical output constraints» (Benua, 1995:122). The full parsing of the base can be also modified by the interaction of faithfulness and unmarkedness constraints, as in Daniel. In spite of the fact that 12 CatWPL 6, 1998 Teresa Cabré 5. For a proposal of the lexical stress asignment in Catalan in OT, see T. Vallverdú (1997). CatWPL 6 007-022 21/10/99 12:44 Página 12
-el is the constituent on which the Anchor-Position constraints operate, the requirements on the Prosodic Word force a disyllabic truncated form. 4. Analysis In Catalan, there are a reduced number of very common words whose output forms are monomoraic (vi ‘wine’, pa ‘bread’, bé ‘good’, por ‘fear’, dur ‘strong’, etc). Starting from the general truncated forms in (3), we can state that Catalan truncated words are minimally bimoraic and maximally bisyllabic. Truncated forms are unmarked prosodic words, so Foot Binarity constraint must be observed. Thus, a low position of Max-BT in the ranking allows the truncated hypochoristics to not copy the whole base. FtBin: Feet are binary at the moraic level or at the syllable level. Max-BT: Each element of the base form has a corresponding element in the truncated form. The tableaux in (6) illustrate two examples of the first group. As is evident, each violation in MAX-BT corresponds to a deletion of a syllable. Some of the examples of this first group display important violations of BTIndentity. We have said that the initial flap of the stressed foot cannot be kept in initial position of the hypochoristic form, but cannot be strengthened either. On the other hand, unmarked onset conditions on truncated outputs force the initial onset position to be filled.6Even though an underlying flap becomes a trill in some syllabic positions, the faithfulness relation is not preserved because it deals with a correspondence relation between two outputs. The best way to be faithful to the Dialectal Variation in Truncated Words in Catalan CatWPL 6, 1998 13 6. It is possible to find a few disyllabic hypochoristics without initial onset: Ita (from Margarita) and Ia (from Maria), but Mita and Mia are more common. Other possible hypochoristics like Oni (from Jeroni) or El (from Daniel) are completely unacceptable. (6) input FtBin Anchor-Pos Anchor-Pos Anchor-Pos MaxBT En.ri.que.ta BT (Ft, head) BT (Ft, ini) BT (Ft, fin) ☞ Que.ta ** En.ri *! * * ** Ri.que.ta *! * * FtBin Anchor-Pos Anchor-Pos Anchor-Pos Max-BT Le.o.pol.do BT (Ft, head) BT (Ft, ini) BT (Ft, fin) ☞ Pol.do ** Le.o *! * * ** O.pol.do *! * * CatWPL 6 007-022 21/10/99 12:44 Página 13
base is by taking one of the unmarked available consonants (there is no trill in the base form), preferably the initial consonant of the base name that identifies it (Kenstowicz, 1996). Thus, nasals and stops are the favoured consonants, while fricatives are marked. In order to restrict the amount of constraints in the ranking we have included different types of faithfulness under BT-Faith, as specified below. Two examples of this are shown in (7). BT-Faith: There must be structural and segmental faithfulness between corresponding outputs. No deletion, no insertion. I-Onset: Truncated forms are required to have initial onset. The violations of BT-Faith mean different things: the initial trill segment of *Roni and *Rita is not present in the corresponding base form, whereas *Oni shows that the initial onset has been deleted. We have seen before too that there are a set of truncated words in the first group that violate faithfulness in another way: the bases have complex onsets and they are systematically simplified in a glide or consonant. The reason is the domination of BT-Faith by a wellformedness condition on syllabic structure that emerges in this kind of words. No C-Onset: Complex Onsets are forbidden in truncated forms. We must say about these cases that the simplification in a glide or consonant depends on the consonant featural characterization: nasal plus glide is reduced to a nasal (Tona), whereas other clusters are simplified into a glide (Coia), unless an adjacent vowel with the same articulatory features blocks it (Cila from Cecília).7 The tableaux in (8) show two examples of this phenomenon: 14 CatWPL 6, 1998 Teresa Cabré 7. These solutions coincide with the evolution of a universal phenomenon of palatalization (Bhat 1978). Nowadays they can display another form like Lali or Cori. Nevertheless, it is possible to find a few hypochoristics with no onset simplification: Xènia (from Eugènia) and Sèbio (from Eusebio). Initial palatalization is another phenomenon that Catalan truncation exhibits: Xènia, Xar, Ximo. (7): Je.ro.ni I-Onset BT-Faith ☞ No.ni Ro.ni *! O.ni * *! Mar.ga.ri.ta I-Onset BT-Faith ☞Mi.ta Ri.ta *! CatWPL 6 007-022 21/10/99 12:44 Página 14
At this point of the analysis we cannot provide a strict ranking of the proposed conditions. We know that hypochoristics must observe wellformedness constraints like FtBin, I-Onset and No C-Onset; Anchor-Pos (Ft) preserves the identity with the head and the edges of the base lexical stress foot; BT-Faith can be violated under the wellformedness constraints. 5. Cases from final syllable stressed bases The second group of truncated forms is certainly the most interesting one. All bases finish in a heavy syllable, the prosodic constituent that holds the lexical stress. This base constituent becomes a Prosodic Word with a leftheaded disyllabic shape in Valencian, whereas in Eastern Catalan one or two syllables are taken from the base under requirements of faithfulness and wellformedness in conflict. Let us consider the first monosyllabic cases as opposed to disyllabic ones. Monosyllabic truncates from disyllabic bases are the single possibility for hypochorization to occur. The tableaux in (9) show two examples: More striking is the mess set of possibilities and ungrammaticalities permitted by the longer bases. Isabel, Joaquim, Segimon or Meritxell can only yield monosyllabic truncated forms: Bel, Quim, Mon, Txell (*Sabel, *Aquim, *Gimon, *Mitxell are completely unacceptable), whereas Bartomeu, Montserrat or Baltasar can yield Tomeu, Serrat, Tasar and also Meu, Rat, Xar. Finally, other bases like Daniel or Sebastià can only yield disyllabic hypochoristics (Niel, Tià), but this is due to the dominance of wellformedness constraints, as we will see later. Dialectal Variation in Truncated Words in Catalan CatWPL 6, 1998 15 (8): Eu.là.lia No C-Onset BT-Faith ☞La.ia * La.lia *! An.tò.nia No C-Onset BT-Faith ☞To.na * To.nia *! (9): FtBin Anchor-Pos Anchor-Pos Anchor-Pos Max-BT Mi.quel BT (Ft, head) BT (Ft, ini) BT (Ft, fin) ☞Quel * Mi.quel *! FtBin Anchor-Pos Anchor-Pos Anchor-Pos Max-BT Re.mei BT (Ft, head) BT (Ft, ini) BT (Ft, fin) ☞Mei * Re.mei *! CatWPL 6 007-022 21/10/99 12:44 Página 15
Kenstowicz, Michael (1996). «Base-Identity and Uniform Exponence: Alternatives to Ciclicity». In: Durand, J.; Lasks, B. (eds). Current Trends in Phonology: Models and Methods. Manchester: European Studies Research Institute, University of Salford, p. 363-393. Nelson, Nicole (1998). Mixed Alignment Effects in French Hypochoristic Formation. Ms. University of Massachusetts Amherst. McCarthy, John (1998). «Faithfulness and Prosodic Circumscription». Ms., U. of Massachusetts, Amherts. McCarthy, John; Prince, Alan M. (1994). «The emergence of the unmarked». Ms., University of Massachusetts, Amherst and Rutgers University. — (1995). «Faithfulness and reduplicative identity». Beckman, J. N. et al. (eds.). Papers in Optimality Theory. U. of Massachusetts Occasional Papers, 18, U. Mass. Amherst, p. 249-384. Prince, Alan S.; Smolensky, Paul (1993). «Optimality Theory: Constraint Interaction in Generative Grammar». Ms., Rutgers Un. and Un. of Colorado, Boulder. Vallverdú, Teresa (1997). On stress assignment in Catalan. Master degree dissertation, Universitat Autònoma de Barcelona. 22 CatWPL 6, 1998 Teresa Cabré CatWPL 6 007-022 21/10/99 12:44 Página 22