Weak Prepositions in Tortosan Catalan : alternation of Prepositions, Allomorphy or Phonological Process

CatWPL 511. 1996 29-66 Weak Prepositions in Tortosan Catalan: Alternation of Prepositions, Allomorphy or Phonological Process* Esteve Clua Universitat Pompeu Fabra. Departament de Traducció i Filologia. u La Rambla, 30-32. 08002 Barcelona (Catalonia) Spain e-mail: clua-esteveC3trad.upf.e~ Received: March 25th 1996 Accepted: July 10th 1996 Abstract In this paper, starting from a brief analysis of weak prepositions in Tortosan. I try to show that the preposition a is the only one appearing in adjunct and indirect object PPs. This forces us to account for the form Can] which this preposition takes in certain contexts: specifically when it is followed by a vowel initial determiner or pronoun. After considering the alternative of accounting for the relation between the preposition a and the form [an] either in terms of prepositional allomorphy or in terms of epenthesis of the coronal nasal, I have chosen the latter option, since it allows us to explain similar processes affecting other prepositions (contra 'against', cap a 'towards' and per a 'for') and the privative prefix a. Optima1 syllabification (CV syllable), which is a preferred but not obligatory condition in Catalan, would be, in its formulation as the Onset Principle, the trigger for this process of consonantic epenthesis, which essentially affects the domain [clitic+word] when the former is a determiner. Key words: morphophonology, prepositions, Catalan dialects. Resum. Les preposicions febles del tortosí: alternanga de preposicions, al.lornorfia o procés fonoldgic En aquest article, i a partir d'una anhlisi breu de les preposicions febles del dialecte tortosí, intento demostrar que la preposició a és l'única que apareix en els complements circumstancials i en els indirectes. Aixb implica explicar la forma [an] que realitza l'esmentada preposició en certs contextos: concretament quan va seguida d'un determinant o d'un pronom comeqats en vocal. Davant la doble possibilitat d'explicar la relació entre la preposició a i la forma [an]: al.lomorfia preposicional i epbntesi de la vocal nasal, m'he decantat per aquesta segona via, ja que permet d'explicar processos similars que afecten altres preposicions (contra, cap a i per a) i el prefix privatiu a. La sil.labificaciÓ bptima (sil.laba CV) -preferent, perb no obligatbria en This is a revised and reduced version of the Research Paper I submitted for the degree of Master in Linguistics at the Universitat Autbnoma de Barcelona. I want to thank the members of the evaluating comrnittee, Joan Mascaro, Teresa Cabré and Eulalia Bonet, for their comments, that have indeed helped in improving the original paper. 30 CatWPL 511, 1996 Esteve Clua catali-, mitjan~ant l'Onset Principle, seria la causa d'aquest procés d'epkntesi consonintica, que afecta bisicament l'imbit [clític+mot] quant el primer és un determinant. Paraules clau: morfofonologia, preposicions, dialectes Catalans. I Table of Contents I 1. Introduction 5. The Tortosan Prepositional System 2. The Prepositions a and en 6. Description of the Phenomenon in Place PPs 7. Explanatory Motivation 3. The Form In] in Indirect Objects 8. Conclusions 4. Locative and I0 Prepositions in Tortosan. Preposition Alternation or Phonological Variation 1. Introduction In the description of the uses of the Catalan weak prepositions a, en and amb,' it is often more or less explicitly assumed that phonetic factors should explain their syntactic distribution in different contexts. My linguistic intuition has always lead me to think that very often it was not a case of phonically conditioned alternation of distinct prepositions, but rather one of prepositional allomorphy or phonological process inducing a phonetic change in one preposition (a). In this paper I try to motivate this descriptive option by analyzing the Tortosan prepositional ~~stem.~ To this end I analyze the main prepositional uses, especially those relevant for singling out and describing the variation shown by the preposition a. After describing the phenomenon, I try to provide an explanatory account in terms of general principles constraining the phonology of natural languages. Treating this phenomenon as a phonological process -consonant epenthesis-, triggered by the application of a general principle, allows us to explain the variation observed with other prepositions (contra 'against', per a 'for', and cap a 'toward'); furthermore, it can contribute to a better understanding of the application and validity of these principles. 1. According to prescriptive grammars a and en are roughly in complementary distribution in their locative (inessivelallative) use, while in free distribution in other distinct uses (dative and govemed); and amb is a completely unrelated comitative preposition. Dialectally and colloquially, however, the facts are quite different, as we will see. 2. Tortosan is a Catalan dialect in the South of Catalonia, belonging to the Westem dialectal area, and neighboring the Valencian dialects, with which it shares some features. Western dialects are standardly divided into Northwestern (West of Catalonia) and Valencian. Eastern dialects are divided into Central (East of Catalonia), Northeastern (Roussillon) and Balearic. Tortosan has only two of these prepositions, a and en, and the present study will therefore only be concerned with these. All the prepositional uses which are expressed by amb in standard Catalan are expressed in Tortosan, like in Valencian, by en. CatWPL 511, 1996 31 I believe that a descriptive and explanatory account of these facts can contribute to shedding some light into the synchronic description of weak prepositions of the whole Catalan linguistic area. The existing studies on the Catalan prepositions a, en and amb quite often talk about confusion, interference or alternation. There are some basic reasons that somehow justify this situation. On the one hand, even if these elements enjoy a significant degree of syntactic and semantic individuality in many cases, this does not seem to be the case in others (locative PPs); furthermore, the fact that all three prepositions are pronominalized with the same clitic, hi (roughly equivalent to the French clitic y), would point in the same direction. In this vein, Ferrater (1981:32) treats these prepositions as combinatory variants of the same word. On the other hand, their phonic nature can explain the alternations and confusions in some cases. Firstly, we are dealing with unstressed, monosyllabic clitic words; even monosegmental ones in the case of the preposition a. Furthermore, their vowel is always the unmarked vowel: [a] in Eastern dialects ([a],[an] and [amb]), [a] and [e] in Western dialects ([a], [en] and [amb]);3 which implies that it can be very easily affected by phonological processes such as vowel e~ision.~ And, finally, the consonanta1 element is, in one of the cases, a coronal nasal, and can therefore be affected by processes of assimilation of point of articulation of nasals. In most of the work on these prepositions, reference is made to the influence of the phonetic context in the selection of the preposition. In Perera (1986: 88) this view is synthesized as follows: As a result of this phonetic weakening, nowadays in most of the Catalan linguistic domain the use of these three forms or of their numerous dialectal variants depends more on the phonetic context where they appear -specifically the initial sound of the following wordthan on the syntactic function they take part in, or on their etymological meaning. Yet, the above-mentioned confusions and interferences can, in my view, very often be explained in terms of general phonological processes or constraints. Provided, of course, that we start from a strictly synchronic point of view of the phenomenon and we make use of the basic tenets in recent phonological theory. The analysis of these prepositions in the Tortosan dialect can, in my opinion, be relevant and clarifying at least for some aspects that constitute an important source of confusion in the study of other dialects. And this is so for three basic reasons. First, in this dialect, like in the rest of Western dialects, the process of unstressed vowel reduction is partial, i.e., it only affects the feature [+ATRI [Advanced Tongue Root], and thus there is vowel distinctness between the prepo3. With North-Western dialects it is not as clear as with the Eastern ones that there is one single unmarked vowel segment. In Palmada (1991:91) it is argued that the status of unmarked is shared by the segments [a] and [e]. 4. See Mascaró (1984, 1989), Wheeler (1979). 34 CatWPL 511, 1996 Esteve Clua What, in my view, does not seem to be so reasonable is that later studies, intended to be descriptive and synchronic, have started from Fabra's analysis, in view of the fact that this analysis, whose ultimate goa1 was clearly prescriptive, had made use of diachronic criteria in an important respect. 2.1. Semantic Differentiation or Phonetically Conditioned Differentiation With respect to Fabra's analysis of the prepositions a and en in locative PPs, which is, directly or indirectly, the starting point of almost all later studies, basically two positions have been adopted. Most scholars that have taken a stand on the issue accept the view that two prepositions are involved. And they attribute the choice of one or the other only to the following phonic context, with no consideration of any syntactic-semantic distinction between the two. From a synchronic point of view, if there is no syntactic-semantic difference between a and en, and the only criteria to choose one preposition or the other are of a phonic kind, perhaps one should contemplate the possibility that we are actually dealing with one single preposition with two (or more) phonetically conditioned allomorphs. We come back to this point later. Among the studies that would take this view (linking the prepositional variation to the phonic context), we can point out the following: Calveras (1928), Gulsoy (1965), Sola (1977), Cabré (1980), Segarra (1985), Mier (1986), Palet (1987), and Perera (1987).~ The other position, rather in the minority, tries to account for the alternation of prepositions in terms of semantic criteria. It is attested in the studies by Camprubi (1984) and Sancho (1994). I think it is in principle a more coherent position than the preceding one, since if there is an alternation between prepositions we would naturally expect to find some kind of syntactic or semantic contrast, at least from a synchronic point of view. Sancho's study, which is the largest and most important of the two, analyzes the Catalan prepositional system from the viewpoint of cognitive grammar. From my perspective, and for what concerns the present study, it suffers from an important methodological problem: it is a study on ccthe Catalan prepositional system as proposed by the normative grammar ... >> which starts ($rom a corpus based on Eastern Catalan texts of various kind~>>.~ 5. Perhaps it should be pointed out that the studies by Calveras, Gulsoy, Perera and Segarra are diachronic and thus considerably depart from the basic tenets I embrace here. Instead, Cabré's (1980) study, as we will see later, is quite different from the others, as it seriously considers the option I defend here. I.e., while the other studies, following Fabra, view the phenomenon as a phonically conditioned change of prepositions, Cabré points out that it might be a case of phonetic fluctuation of the preposition. 6. The author justifies this methodological choice on the grounds that a corpus based on data of other kinds would involve a very high degree of complexity and also because: aOur interest is to see ifa cognitive model can explain how the Catalan prepositional systenz (or systems) works, with specijic attention to the degree of motivation for the various uses of the main Catalan prepositions.> Weak Prepositions in Tortosan Catalan CatWPL 511, 1996 35 Sancho accounts for the alternation between the prepositions a and en in spatial terms, mainly for inessive uses, on the basis of semantic considerations: a) The preposition 'a' expresses pure location, regardless of the size of the LM (landmark), which is conceived punctualiy. For pure location it is required that the LM be properly present in the presuppositions shared by speaker and hearer; hence 'a' is used with place names, the definite article and nouns associated to linguistic routines (301). (301) Joan és a Valbncia / al jardí / a casa. Joan is a Valencia / a-the garden / a house 'Joan is in Valencialin the gardenlat home.' b) The preposition 'en' expresses location in relation to the structure outerlinner part of the LM, i.e., as it expresses location it establishes a relation containerl contained. This conceptualization arises whenever the exact place of the LM is not sufficiently known, namely with indefinite articles and demonstratives (302). (302) Joan viu en una ciutat / en aquesta ciutat. Joan lives en a town / en this town (Sancho 1994:125) Apart from what I have pointed out above concerning methodological issues, there is a couple of remarks to be made to this study. On the one hand, the contrast concerning [+ definiteness] between the article and the demonstrative does not seem very clear in contexts such as the ones in (2):7 (3) a. Viu a la casa del seu pare. live.3.q a the house of-the his father 'She lives in his father's house.' b. Viu en aquesta casa del seu pare. live.3.q en this house of-the his father 'She lives in this house of his father's.' On the other hand, such an analysis would not account for the facts in Tortosan, since in this dialect there is one single preposition independently of the element following it: (4) a. Viu a la casa de dalt. live.3.q a the house of above 'She lives in the house above.' b. Viu a[n] esta casa de dalt live.3.sg a[n] this house of above 'She lives in this house above.' 7. In this sense, the author observes: aIn the case of demonstratives, however, one should not discard the possibility that the avoidance of the hiatus has contributed to the usage of 'en',,. 36 CatWPL 511, 1996 Esteve Clua The account of these prepositions by Badia i Margarit (1994:200,215) is worth a special mention, since it would not be fully ascribable to either of the preceding options. This linguist in fact adduces phonetic reasons which determine the form of the preposition a, but, as we will see, he does not tie the use of the locative prepositions to the phonic context in a direct way. In short, he considers the possibility that the form [an] in locatives not only may be related to the preposition en, as in most grammars, but it also may be a realization variant of the preposition a, determined by antihiatic reasons. 2.2 Prepositional Allomorphy As I briefly mentioned in the preceding section, from a synchronic perspective the view that the alternation between a and en is linked to the kinds of following phonic context seems to inevitably lead to treating these two prepositions as allomorphic variants of one single preposition. At least in the case of locative PPs. This is the option taken by Ferrater (1981:32). According to him, the prepositions a, en and amb are combinatory variants of the same word; and the selection of one of the variants is determined by whether the following word begins with a vocalic or consonanta1 segment. His proposal, however, goes beyond the field of locatives to cover the various meanings involved in these prepositions. More than a proposal, it is a short (but lucid) outline of the prepositional system. Its shortness is probably the reason why it is only a poorly motivated outline, in the sense that it does not sufficiently explain the phonic conditioning, and it only mentions as counter-examples the case of toponyms that, in spite of being vowel-initial, require the preposition a. Furthermore, in other prepositional uses (the so-called governed PPs, for instance), it is not clear at all that there should be no syntactic-semantic distinction between the three prepositions. Specifically, in these cases we can indistinc~ly find a and en before a word beginning with a vowel or a consonant: (5) a. Joan no s'ha acostumat a treballar de nit. Joan not rep-has accustomed a to-work of night 'Joan hasn't got used to work by night.' b. Joan no s'ha acostumat a aguantar el soroll. Joan not rep-has accustomed a to-stand the noise 'Joan hasn't got used to stand the noise.' c. Joan no s'ha acostumat al ritme de vida de la ciutat. Joan not rep-has accustomed a-the rhythm of life of the town 'Joan hasn't got used to the urban way of life.' d. Els estudiants només pensen en les notes. the students only think en the grades 'Students only think about their grades.' e. Els estudiants només pensen en aquell examen. the students only think en that exam 'The students only think about that exam.' Weak Prepositions in Tortosan Catalan CatWPL 511. 1996 37 Yet I think this is an analysis to bear in mind, at least for the case of locatives. 3. The Form [an] in Indirect Objects One of the prepositional uses where most authors agree in pointing out some phonic factor for the form of the prepositions (or rather the preposition) is that of indirect objects (10s): (6) a. Ho he donat a ell. it-have.l.sg given a him 'I have given it to him.' b. Ho he dit a algú. it-have.l.sg said a someone 'I have said it to someone.' c. Ho he portat a aquest noi. it-have. l.sg carried a this boy 'I have carried it to this boy.' d. T'ho he donat a tu. you.dat-it-have. l.sg given a you 'I have given it to you.' The account of this an forn in 10s ([an] in Eastern dialects and [an] in Western dialects) differs considerably, as we can see, from the prevailing analysis for the identical forns in place PPs. As I already mentioned, Badia (1994) considers the possibility that the form an may be a realization variant of the preposition a, in which the nasal segment would play the role of antihiatic banier so as to prevent the preposition to be confused with the vowel in the following word, especially ifthe latter is unstressed too. According to this linguist, while the un of place PPs allows for two possible accounts (the preposition en from Old Catalan and the antihiatic variant of the preposition a), the sequence un in 10s can only be related to the preposition a, the only preposition introducing this kind of c~m~lement.~ The analysis of prepositional uses in 10s reveals a clear asymmetry in the accounts that have been given for these elements, an asymrnetry already present in the formulation by Cabré (1980: 14-15). In the case of place PPs, since the ancient language seemed to use both the prepositions a i en, the form un is either related to the latter, or -as in Badiait is left undecided whether it should be related to en or to a (if the latter, for antihiatic reasons). In the case of IOs, instead, it is mainly related to a, because it is hard to link it to the evolutive outcome of the Latin preposition in: 8. Even if this is not explicitly stated, as it is in the case of locatives, it is the only conclusion that can . be drawn. Other references to this antihiatic element are: Barnils (1921:60), Solh (1977:38), Mier (1986:88), and Recasens (1991:262-263). 38 CatWPL 511, 1996 Esteve Clua (7) LOCATIVES. a) Estudia en aquesta facultat. [aril study.3.sg en this faculty 'Ske studies in this school.' b) Va pujar en aquella torre. ran] pst.3.sg climb en that tower 'Ske climbed that tower.' (8) 10s. a) Li ha enviat flors a ella. [an] dat.3.sg-has sent flowers a her 'Ske has sent her flowers.' Thus, from a strictly synchronic perspective, and in view of what happens with IOs, we could assume that place PPs, both locative and allative, only involve the preposition a, which receives a consonantic support [n] before certain vocalic contexts. I think that this option offers a more comprehensive account of these prepositions and enables us to overcome the above-mentioned contradiction. 4. Locative and I0 Prepositions in Tortosan. Preposition Alternation or Phonological Variation The data from this dialect constitute an important piece of support for the hypothesis I put forth above. As I already pointed out, the fact that this dialect has the unstressed vowel system in (9) makes the phonic differentiation between the prepositions a and en not exclusively rely in the presence of the consonantic segment, but also in the timbre difference in the vowel segment: a [a], en [en]: In Massip (1979:162), which, together with Calveras (1928), constitute the only references to these prepositions in Tortosan, it is remarked that the preposition a before demonstratives and in some other cases (the indefinite algun, etc.) is strengthened with the consonant n in Tortosan. As for place PPs, the examples in (10) show us that there are two possibilities, [a] and [an]: (10) a. Ha pujat en aquell pis. has climbed en that storey 'Ske has gone up that storey.' b. Ha pujat al pis de dalt. [al has climbed a-the storey of above 'Ske has gone upstairs.' c. Ho trobaras a la caixa. it-find.fut.2.q a the box 'You'll find it in the box.' Weak Prepositions in Tortosan Catalan CatWPL 511, 1996 39 (10) d. Ho trobaras en esta caixa. it-find. fut.2.sg en this box 'You'll find it in this box.' e. Viuen a Amposta. [al live.3.pl a Amposta 'They live in Amposta.' f. Viuen a Ambrica. [al live.3.pl a America 'They live in America.' g. Viuen en algun país llunya. [analyÚn] live.3.pl en some country distant 'They live in some faraway country.' It seems initially plausible that the alternation between these two forms is phonically driven. Except for the cases (e) and (0, where the NP is a toponym, in all other cases the segment [a] occurs before consonanta1 segments, while the sequence [an] is found before vocalic segments. As for which preposition underlies these phonetic realizations, I think that it can only be the preposition a. If it was the preposition en, what we would find is the forn [en], as illustrated in the examples in (1 I).~ (1 1) a. Sempre pensa en tu. always thinks en you 'Slhe's always thinking of you.' b. Sempre pensa en estes coses. [en] always thinks en these things 'Slhe's always thinking about these things.' c. Ho faré en tres dies. [en] it-do.fut. 1. sg en three days '1'11 do it in three days.' d. Se'ls menjava de dos en dos. [en] rep-them-eat.3.sg of two en two 'Slhe was eating them two at a time.' e. Ho ha escrit en guix. it-has written en chalk 'Slhe wrote it with chalk.' The preposition in 10s shows the same possibilities of realization, [a] and [an]: (12) a. Li hovan donar a aquella altra. [anakélla] dat.3.sg-it-pst.3.pl give a that other 'They gave it to that other one.' 9. The process of reduction of pre-stressed [e], which changes mid front vowels to [a] in NorthWestern dialects, does not occur in these contexts in Tortosan, as has sometimes been suggested for these dialects. See Sol& (1977). 46 CatWPL 511, 1996 Esteve Clua Tortosan, like in Valencian, this preposition is used in cases corresponding to amb 'with' in the other Catalan dialects.15 The examples in (18) further support the need to assign the form [an] to the preposition a, and not to en, since the latter is always realized as [en] : (18) a. Sempre ha treballat en ganes. [en] always has worked en wish 'She has always worked willingly.' b. Sortirem en aquells amics. [en] go-out.fut. l.pl en those friends 'We'll go out with those friends.' c. Ho va escriure tot en esta ploma. [en] it-pst.3.sg write all en this pen 'Sihe wrote everything with this pen.' d. Sempre ha viatjat en cotxe 1 en tren 1 en avió. [en] always has travelled en car / en train 1 en plane 'She has always travelled by car / by train 1 by plain.' e. S'han pansit en la calor, les flors. [en] rep-have.3.pl withered en the heat the flowers 'They withered from heat, these flowers.' 5.5. Governed PPs Like in the rest of dialects, Tortosan prepositional governed PPs can be introduced by a and by en (beside other prepositions): (19) a. No crec en aixb. not believe.l.sg en this 'I don't believe in this.' b. No s'ha acostumat a aixo. not rejl-has accustomed a this 'She hasn't got used to this.' c. Estic d'acord en allb. am of-accord en that 'I agree with that.' d. Hi tenia dret, a allb. there-have.pst.3.q right a that 'Slhe had the right to that.' e. Té molt d'interbs en estes coses. [ en] has a-lot of-interest en these things 'Sihe has a lot of interest in these things.' 15. For a diachronic account of this phenornenon, see Calveras (1928) and Gulsoy (1965). Weak Prepositions in Tortosan Catalan CatWPL 511, 1996 47 (19) f. Sempre pensen en aquella histbria. [en] always think.3.pl en that story 'They are always thinking about that story.' g. Ves en compte en l'alcohol. [en] go.imp.sg en account en the-alcohol 'Be careful with alcohol.' h. Sempre t'exposes a algun risc. always you-expose.2.sg a some risk 'You're always exposed to some risk.' i. S'han dedicat a apagar focs. [a~a~ál rejl-have.3.pl devoted a extinguish fires 'They have devoted themselves to extinguishing fires.' Examples (19.b,d) show us two further cases where the form [an] appears: the demonstrative pronouns aixi, 'this' and alld 'that'. If we compare them with (19.a,c), respectively, we can clearly see that they are only related to the preposition a. In (19.i) we have another context where the preposition, even if followed by a vowelinitial element -specifically an infinitive- , does not show the form [an]. 5.6. The Direct Object The use of the preposition a with these complements in the Tortosan dialect is very similar to its use in Spanish, narnely, the preposition appears before [+human] direct objects: (20) a. Ja els he buscat, a ells. [anéds] already them-have.l.sg[looked-for a them 'I already looked for them.' b. He buscat a estos amics tot el dia. [anéstos] have. I.sg looked-for a these friends all the day 'I have been looking for these friends the whole day.' c. Tota la vida ha buscat amicslpares, i mira All the life has looked-for friendstparents and look com es troba ara. how re$-finds now 'All hislher life she's been looking for friendslparents, and look how she is now.' d. Els policies perseguien a aquells lladres. [anakéds] the policemen chased a those thieves 'The policemen chased those thieves.' e. Els mossos devien seguir a algú. [analy;] the policemen must.pst.3.pl follow a someone 'The policemen must have been following someone.' 48 CatWPL 511, 1996 Esteve Clua (20) f. Seguien a Agustina pertot arreu. [al follow.3.pl a Agustina all over 'They were following Agustina all over.' The only remarkable point in this case is the absence of the preposition with [+human] NPs lacking a determiner (20.c) and the absence of the segment [n] before proper names (20.0. The data about time and manner PPs, governed PPs and direct objects lead us to the conclusion that the appearance of the nasal segment [n] in certain vocalic contexts after the preposition a is not an isolated phenomenon concerning only 10s and locatives, but rather it covers all the possibilities of appearance of this preposition. And this is in fact what one should expect if it were a phonologically conditioned phenomenon. 5.7. Other Prepositions Let us now see whether the analysis of other prepositions, whose final segment is the same as the single segment in the preposition a, can help us understand the phenomenon. The prepositions I will analyze are: per a 'for', cap a 'toward', contra 'against' and com a 'as',16 namely, all those which have a segment [a] in final en position and, therefore, are relevant for the phenomenon under exarnination. The analysis of these prepositions, in which the segment [n] appears too, leads us to the conclusion that we are dealing with a fact which is not confined to the preposition a, and, consequently, seems to require a phonological explanation, rather than a morphological one. The Preposition per a 'for'. Unlike what happens in dialects such as Central, where the distinction between per 'by' and per a 'for' has been lost and per has taken over both over bothe of them, in Tortosan the distinction is entirely alive and functional. In (21) I present the realizations of the prepositionper a in the sort of context I am analyzing: (21) a. Fa molts anys que treballa per a ells. [peranélls] does many years that work.3.sg per a them 'Slhe has been working for them for many years' b. Per a estes persones tot és corrupció. [peranéstes] per a these people all is cormption 'According to these people, everything is corruption.' c. Treballa per a algú que no coneixem. [peralyu] l7 work.3.sgper a someone that not know.I.pl 'She works for someone we don't know.' 16. I will not consider the preposition sota 'under', because it is not generally used in Tortosan. 17. In these cases, like in Valencian, we also find the realization [palyÚ]. The liveliness of this form is not, however, as strong as in Valencian,and, furthermore, it depends very much on the variable age of the speaker. Older people use it more profusely. Weak Prepositions in Tortosan Catalan CatWPL 511, 1996 49 (21) d. Al final es va desdir. Per a aixb no [perajj;] In-the end rejl-pst.3.sg unsay per a this not calia tant de soroll. was-needed so-much of noise 'In the end ske went back on his word. It had been much ado about nothing.' e. Verema per a un d'aquells. [peranÚn] pick-grape.3.q per a one of-those 'Ske's working in the grape harvest for one of those.' f. Ho ha portat per a aquells xiquets. it-has carried per a those boys 'Ske brought it for those boys.' g. Ha vingut per a anar a escola. has come per a go to school 'Slhe has come to go to school.' The Preposition cap a 'towards'. This preposition, which is used with verbs of movement to express orientation, has the following uses: (22) a. Es dirigia cap al museu. [kapal refl-address.pst.3.sg cap a-the museum 'Slhe was heading for the museum.' b. Se'n va anar cap a Astúries. [kapal rejl-cl-pst.3.q go cap a Astúries 'Slhe left for Astúries' c. Ha marxat cap a escola. has left cap a school 'Ske has left for school.' d. Si vas cap a esta vall, en trobaras. [kapanistal if go.2.sg cap a this valley thereof-find.fut.2.sg 'If you go towards this valley, you'll find some.' e. Anava cap a ella amb una cara! [kapaneda] go.pst.3.q cap a her with a face 'Ske was approaching her with such a sulky look!' f. Corrien cap a aquella banda d'alli. [kapakéda] run.pst.3.pl cap a that side . of-there 'They were running in that direction.' g. No sé si anar cap aquí o cap allh. [kapaki]/[kapadá] not know. 1 .sg if to-go cap a-here or cap a-there 'I don't know whether to go over here or over there.' 18. On the use of the preposition per a in Tortosan, see Massip (1979). 50 CatWPL 511, 1996 Esteve Clua The Preposition contra 'against'. The examples of this preposition, like those of the other two above, show a difference in behavior that we had not observed in the other cases: (23) a. Contra aixb /contra allb, no hi pots [kontraj~~/kontraA6] contra this / contra that not there-can.2.sg fer res. do nothing 'Against thislthat, you can't do anything.' b. La setmana passada van jugar contra Alemanya. [kontra] The week past pst.3.pl play contra Germany 'Last week they played against Germany.' c. No es pot anar contra animals com estos. [kontra] not refl-can.3.sg go contra animals like these 'You can't fight animals like these.' d. El vent batia contra aquells vidres. [kontrakélis] the wind strike.pst.3.sg contra those glasses 'The wind was blowing against those glasses.' e. Es va revoltar contra ells. refl-pst.3.sg revolt contra them 'Sthe revolted against them.' f. És un bon remei contra esta malaltia. [kontranésta] is a good remedy contra this illness 'It's a good remedy against this illness.' As can be seen by comparison with the preposition a, in addition to the cases where the preposition is followed by a vowel-initial [+NI (proper name or a common noun), with these prepositions the segment [n] fails to appear before: aquell(inflection) 'thatlthose', algú 'somebody' algun(inflecti0n) 'some', aixd 'this' and all6 'that'. From a phonic point of view, what these elements share is the fact that they begin by the segment [a], unlike the ones that trigger [n], which begin by [e]. It seems that in these constructions the context required in order for the [n] to appear is [a e] (+[anc]), while the context [a a] would give as a result the simplification of the two identical adjacent segments: [a]. The difference in behavior with respect to the realizations of the preposition a is obvious, since in the latter case [n] also appears in the context [a a] with the elements aquell(inflection), aixd, alld, algú, algun(inf1ection). (24) a. Vivien a aquell castell. live.pst.3.pl a that castle 'They lived in that castle.' b. Deu haver anat a alguna part. must.3.q have gone a some part 'She must have gone somewhere.' Weak Prepositions in Tortosan Catalan CatWPL 511, 1996 51 (24) c. S'exposava a aixb. refZ-expose.pst.3.sg a this 'Slhe was running this risk.' The Preposition com a 'as'. At first glance, the behavior of this preposition with respect to the phenomenon at issue seems identical to that of the preceding prepositions: (25) a. Em va ajudar, com a amic que és. [komamik] me-pst.3.sg help com a friend that is 'He helped me as the friend he is.' b. Es va presentar com a aquell amic que [komakék] refl-pst.3.sg introduce com a that friend that els salvaria. them save. cond.3.sg 'He introduced himself as that friend that would save them.' This is, however, a very different case since, as correctly pointed out by Sancho (1994:240-243), this preposition cannot appear with NPs which have a determiner. In these $ases the preposition com 'like' appears instead. Thus, the realization [komakek] in (30.b) is not the result of the simplification of two identical adjacent segments ([a a]), but rather the simple realization of com aquell.19 6. Descrigtion of the Phenomenon To sum up, we can say that in Tortosan the preposition a is the only one appearing in (inessive and allative) place PPs, both in proper place and in figurative pla~e;~O in indirect objects and in direct objects. As for time PPs, this preposition is again the one which is used in most cases; it seems that only time PPs expressing duration are introduced by the preposition en. With verbs subcategorizing govemed PPs, we can either find a or en: in these cases verbal subcategorization determines which of these prepositions must be used. And finally, manner, instrument and company PPs generally take the preposition en. When the preposition a is followed by a vowel-initial element, the consonanta1 segment [n] appears between the preposition and the vowel. It blocks the possibility of syllabic simplification or hiatus. As I have demonstrated, in the case of the prepositions contra, per a and cap a the same phenomenon takes place, with some slight changes. We have seen, however, that this phenomenon is not completely 19. In Sancho's (1994:242) words: *The fact that un 'a(n)', el 'the', aquest 'that' cause a to "drop" is only apparent: rather it simply happens that un, el, aquest convey un individualization of the target object of comparison which is incompatible with the meaning of com a, which is one of total identity. >> 20. As I pointed out above, though, we must bear in mind that PPs with a manner or instrumental interpretation, rather than a place interpretation, are generally expressed by the preposition en. 52 CatWPL 511, 1996 Esteve Clua systematic: it only takes place when the element following the preposition is a determiner or a vowel-initial pronoun: (26) Indefinite article: un, una, uns, unes 'a'(fem)(pl) (not systematic) Demonstratives: este, esta, estos, estes 'this'(fem)(pl) aquell, aquella, aquells, aquelles 'that'(fem)(pl) uix6 'this' all6 'that' Quantifiers: algun, alguna, alguns, algunes 'some'(fem)(pl) un, una, uns, unes 'a'(fem)(pl) (not systematic) algú 's~mebod~'~~ Personal pronouns: ell, ella, ells, elles 'he', 'she', 'they'(fem) In these cases, the appearance of the segment [n] is systematic. Considering that the elements in (26) are the only ones in their classes beginning in a [+syllabic] segment, I think that the phenomenon can be generalized to the determiner category (articles, demonstratives, possessives, numerals and quantifiers) and to the personal, demonstrative and indefinite pronouns.22 In contrast, we never find the segment [n], even if the following element has a [+syllabic] segment in initial position, in the rest of possible contexts, namely, if this element is: (27) A proper name An infinitive A common noun An adverb An adjective 21. In the group of demonstratives we find elements that can be determiners or pronouns (ai& and alld are always pronouns); as for quantifiers, algun(fem)@l) belongs to the determiner category, while algú is an indefinite pronoun; un(fem)@l) can function as a determiner or a pronoun. 22. The definite article in this dialect always begins with the lateral segment [I], since, whenever the syllabification requirements cause the appearance of the morphologically conditioned, [feminine], epenthetic vocalic element ([o]), it is always placed to its right. (i) Lo xiquet no pensa tornar. the boy not thinks come-back 'The boy is not intending to come back.' (ii) Los tarongers s'han gelat. the orange-trees refl-have frozen 'The orange trees have frozen.' Furthermore, the fact that the vowel segment of the preposition occurs before the article renders it completely unnecessary for an epenthetic process motivated by syllabification requirements to take place, since the [I] of the article can syllabify in the coda position dependent on the vocalic nucleus provided by the preposition. As for the other elements in these categories I have not mentioned, all of them have a consonantal initial segment. The only case which, despite its initial vowel, does not trigger [n] is the numeral onze 'eleven': (iii) Ho ha enviat a onze amics. it-has sent a eleven friends 'He sent it to eleven friends.' Weak Prepositions in Tortosan Catalan CatWPL 511, 1996 53 With this state of affairs, we can conceive of two possible options to account for this phenomenon. On the one hand, it could be a case of allomorphy of the preposition a, similar to the one involved in per lalpel 'for the.fem'l'for the.msc'; the number of contexts where the phenomenon does not occur and the predictable lack of strictly phonological reasons to account for them lead, in principle, to such an explanation. On the other hand, we could be dealing with a phonological process of insertion of a coronal nasal in these contexts. In other words, it would be a case of consonanta1 epenthesis. The fact that this segment may appear in other cases would be a sound argument in favor of this option. We have already seen the cases of the prepositions per a, cap a, and contra. There is still another case which is reminiscent of the same phenomenon: the privative prefix a, which is classified as genuinely unstressed (see Fabra (1956:151) and Mascaró (1986:27)): (28) a. amoral, atípic, anormal, acrític, apolític, asimktric, aconfessional, etc. b. analfabet, anharmbnic, analcohblic, anovulatori, anidic, analgbsic, etc. The variation in these forms has been generally treated as a case of allomorphy, but it could be accounted for on similar assumptions similar to the ones I use to account for the preposition a. It is true that cases of allomorphy are fairly common among unstressed prefixes: (29) des, de: deslligar 'untie', desinflar 'deflate', deformar 'deform', deverbal, etc. es, e: escurGar 'shorten', esblanquezt 'discolored', enumerar 'enumerate', etc. trans, tras, tra: transformar 'transform', trasplantar 'transplant', trabucar 'overturn', etc. in, i: inhumd 'inhuman', impossible, innoble 'inoble', irrepetible 'unrepeatable', inacceptable 'unacceptable', etc. And that, this being so, the variation a/an is plausibly nothing but allomorphic variation. If, however, we compare the phonetic contexts where the nasal forms appear with this prefix and those of, for instance, the prefix inli, we can see that the distribution is different. In the former case, the forn [an] only appears before a vowel; in the case of the prefix inli, instead, the form with a nasal can appear before both a vowel and a consonant. Thus, while the forms of this affix are a clear case of allomorphic variation, the case of the negative prefix a seems to be amenable to a phonological ac~ount:~~ (30) in, i: inexacte 'inaccurate', illetrat 'illiterate', inútil 'useless', imparell 'odd', immoral 23. In Mascaró (1995), however, it is suggested that this case could be treated in a way similar to the cases of phonologically conditioned externa1 allomorphy. 54 CatWPL 511. 1996 Esteve Clua (30) a, un: anormal 'abnormal', apbtal 'apetalous', anastigmatic, "ammoral, amoral 7. Explanatory Motivation Concerning the first option, the allomorphic account, Mascaró (1995) presents a very interesting analysis which would provide an explanation for the altemation [a]/[an] as a case of phonologically conditioned externa1 allomorphy of the preposition a. Within Optimality Theory (OT), this phonologist analyzes severa1 cases of allomorphic variation in Romance languages, which show a number of similarities with the phenomenon of variation considered here. The phonological factors which determine the choice of an allomorph in these cases would be related to syllabic markedness and the general preference in natural languages for the unmarked syllabic structure: CV. The second of these options seems, at first glance, the hardest one to justify. It is, however, the most attractive one if we are to attain a higher degree of generalization in the description of the phenomenon. In addition, in the case analyzed here there are further arguments in favor of this account. On the one hand, the phonological context where it takes place is completely uniform: On the other hand, the phonetic alternations it involves are not idiosyncratic. And finally it does not have a reduced lexical scope (a few lexical units), but it rather comprises the set of clitic elements. For all these reasons, I will try to explain the phenomenon as a phonological process. Epenthesis phenomena are quite usual and varied in Catalan dialects. Tortosan is not an exception; the examples in (30) are a sample: (32) a. [askríwre] 'to write' (Cf.: transcriure 'transcribe') [asthp] 'stop sign' (loan from English) [askda] 'school' (Cf. autoscola 'driving school') b. [kÚzes] 'sew.2.sg7 (Cf. surts 'go-out.2.sg9, sents 'hear.2.sg9) [patíses] 'suffer.2.sg' c. [meiriw] 'me-tells' (Cf. m'agrada 'me-pleases') [teiriw] 'you-tells' (Cf. t'agrada 'you-pleases') [seiríw] 'refl-tells' (Cf. s'agrada 'rep-pleases') d. [fuyáses] 'fleeting.fem.pl' (Cf. fuga$, fuga~os 'fleeting.sglmsc.pl') e. [kázos] 'cases' (amiclamics 'friend(s)') [pÉj~os] 'fishes' [astimarlo] 'love. inf-him' (Cf. l'estimo 'him-love.l.sg', estima'l '1ove.imp-him', l[o] vull estimar 'him-want. l.sg love') f. [béndre] 'to sell' (Cf. ven 'sell.3.sg') Weak Prepositions in Tortosan Catalan CatWPL 511, 1996 55 In most of these cases epenthesis seems to apply motivated by phonological reasons: in order to repair ill-formed syllabic structures or to make well-fonned syllabification possible. This is the case with (32.a,c,f). In other cases, such as (32.b,d), its application seems to depend on some sort of morphophonological conditioning (othenvise the person or plural morpheme could not be realized). We also find cases of grammatically conditioned epenthesis, as in the case of the [o] appearing in masculine nominals before the plural morph whenever the stem ends in a [-continuant] strident, or with the 3rd person clitic (examples in (32.e)). And finally, despite the fact that epenthesis is as a mle vocalic, we also find examples of consonanta1 epenthesis: in (32.0 the segment [d] allows the flap be adequately sillabified, since it cannot appear as the only element in the onset when the preceding syllable is closed or when it is in word initial position. The cases in (32.b,d), even if amenable to the above-mentioned morphophonological account, can also be explained by strictly phonological terms. The insertion of the (Western dialect) epenthetic vowel [e] can be explained in terms of wellformedness conditions on syllables, related to the sonority hierarchy in the syllabic coda. We are referring to the SSP (Sonority Sequencing Principle), which requires decreasing sonority for the elements in a syllabic coda starting from the nucleu~.~~ Since in these cases the required decrease in sonority does not occur between the two sibilants, a vocalic segment would be inserted to permit the appropriate syllabification.25 It should be remarked that this outcome ensues whenever the two heteromorphemic sibilants are adjacent in the domain the word. When the boundary between the sibilants is a word boundary, or a clitic-word boundary, the result is usually the merge of the two segments into one: b. [tré.s] [saks] -> [tré.saks] c. [sis] -> [le.sis] This is not, though, the only explanation that has been given to this phenomenon. In Palmada (1991) the appearance of an epenthetic vocalic segment between two sibilants is not explained in terms of wellformedness conditions on syllables, but rather it is accounted for in terms of cooccurrence constraints on this sort of segments, determined by the Obligatory Contour Principle (OCP) (35). According to this author, the OCP is an active principle, in the sense that, besides expressing generalizations concerning an important part of the phonological component, it can block the application of phonological rules which would violate it, and also trigger processes that will override structures not abiding by it. 24. See Kenstowicz (1994:254). 25. For motivation of the rules of insertion or epenthesis in terms of the rules of syllabification see Mascaró (1989:31-61). 62 CatWPL 511, 1996 Esteve Clua different. Among other things, because the requirement of a syllable onset in Catalan is not absolute, but only relative. As for the cases which trigger [n] insertion, they can be sorted as follows: NP detenniners, and personal, demonstrative and indefinite pronouns. Following the nowadays standard view (stemming from Abney 1987) that treats pronouns as (intransitive) determiners we may conclude that the appearance of the epenthetic nasal takes place, in the case of prepositions, when they are adjacent to a vowelinitial deterrniner. This could lead us to assume that there is a stronger structural link between the preposition and the determiner, which would require the appearance of the epentehtic segment, than between the preposition and a head of the category noun, adjective or infiniti~e.~~ The application of the OP in the cases discussed here is closely linked to processes stemrning from vowel contact: breaking a hiatus, vowel elision, diphthongization, etc. There is rather ample consensus that in the possible outcomes of vowel contact different factors play a role: phonological, prosodic, frequency of use, loss of information, etc.39 Now, one factor that groups together all contexts where [n] insertion takes place to the exclusion of all other cases is the high frequency of use of the former. For they are, in short, grammatical units, whose frequency is higher than that of lexical units. At least in oral speech and in a non formal register, the higher frequency of the (a) contexts over the (b) contexts in (46) is obvious: (46) a. Prep [NP Det [N.. .]] b. Prep [NP [NI ... 11 The same can be said about proper names. Their frequency of use as compared to that of determiners or personal, demonstrative or indefinite pronouns is rather low. Besides, in the case of personal names, we must take into account the use of the personal article, which breaks the possibility of vowel contact with the prepo- ~ition.~O Following this reasoning, it can be argued that the preference for CV syllables in Catalan, in accordance with the principles of UG, has developed, among other strategies tending to favor core syllables, a process of consonant epenthesis blocking vowel contact, in contexts of high frequency of use. Moreover, the systematic occurrence of the phenomenon in these contexts, and the fact that it is not restricted to the preposition a are, in my view, decisive for establishing its phonological status. 38. I owe this comment to Jaume Soll. 39. In this connection, see Oliva (1992:157-166) and Recasens (1993). 40. The personal article, which in part of the Catalan dialects (Tortosan among them) is identical to the definite article, is used before a person name. Valencian and adjacent dialects do not have it. In Tortosan, like in the case of pronominal ~Iitics, when syllabification of the underlying form or the article ([I]) requires an epenthetic vowel, it is inserted to the right of the lateral segment. We must bear in mind, however, that the use of this article in Tortosan is far lower than in other dialects. And it is subject to local variation. In most districts it is not used, in others it is used only before feminine person narnes, and in otkers it is used throughout. Weak Pre~ositions in Tortosan Catalan CatWPL 511. 1996 63 8. Conclusions Starting from the analysis of the different uses of the prepositions a and en in the Tortosan dialect, I have tried to show that the preposition appearing in place complements, like the one in IOs, is only the preposition a. The vowel system in tbis dialect enables us to clearly establish this conclusion. This assumption opens the possibility to treat in a uniform way the realizations [an] occurring in locatives and in IOs, as well as in other PPs introduced by a. Furthermore, it characterizes a more coherent prepositional system, since it does not have to link a change of preposition to mere variation of the NP determiner. Upon denying any link between the form [an] and the preposition en, I am forced to try to account for this form in strictly synchronic terms. After considering the alternative of either accounting for this form as a case of allomorphy of the preposition a, or as the result of a phonological process, I have chosen the latter, since it allows us to capture relevant linguistic generalizations and is more explanatory. The case of the prepositions contra, cap a and per a, which show the same segment [n] in some vowel contexts, and possibly the case of privative prefix a (ex.: apolític), can be explained as involving the same phonological process: insertion of a coronal nasal. This is one of the most important arguments in choosing the phonological account over the prepositional allomorphy account for these phenomena: with the latter option the recurrence of the segment [n] in all these forms would come out as mere coincidence. I have considered the possibility that this epenthetic process is triggered by the OCP, a solution parallel to the ones proposed by other authors to account for vowel epenthesis between sibilants. The unexceptionality, on the one hand, of the effects of this principle and, on the other hand, the numerous counterexamples of the phenomenon under study, all of them unaccountable in phonological terms, have led me to abandon this hypothesis. I conclude, then, that consonanta1 epenthesis with these prepositions can be explained on the basis of the Onset Principle. Obviously, this implies accepting the view that this principle is able to trigger phonological processes. This hypothesis, though, is in agreement with the trend that derives epenthesis phenomena from syllabic structure, as an obligatory process. Now, while vowel epenthesis applies because of constraints on well-formed syllabification and is therefore an obligatory process, consonant epenthesis of the sort considered here applies in order to obtain an optima1 syllabic structure, which in the case of Catalan is not obligatory, but preferential: a syllable with a CV on~et.~' The difference in behavior between these two epenthesis processes would be explained by the following contrast: 41. As I mentioned before, Mascaró (1995) makes a similar argument concerning the choice of allomorph in cases of phonologically conditioned external allomorphy. He establishes the following generalization: The allomorph chosen is the one which can be best syllabified with adjacent words to yield un output closer to a core syllable (i.e. CV). With this principle and through the evaluation of different allomorphs according to a ranking of phonological constraints (PARSE-SEG, FILL, ONSET, NO-CODA, etc.), the fittest phonetic variant is chosen, depending on the adjacent context. 64 CatWPL 511. 1996 Esteve Clua (47) a. well-formedlill-formed syllabification -+ obligatory vowel epenthesis b. optimal/acceptable syllabification -+ non-generalized consonant epenthesis As a consequence, the application of this process is not obligatory, and shows a high degree of optionality. It applies rather regularly in the domain [clitic+word], mainly when the clitic is a monophonematic morpheme and in contexts of high frequency of use: before determiners. Finally, I believe that this analysis could be applied without much modification to other dialects which show even stronger variation in their weak prepositions than the dialect described here. In these dialects, though, an explanation in tems of allomorphy would probably be more plausible, at least in those cases where variation takes place in less uniform phonological context^.^^ References Abney, S.P. (1987). The English Noun Phrase in its Sentential Aspect. MIT: PhD.Dissertation. Badia i Margarit, A.M. (1981). ~ramática catalana. 2 vols. Madrid: Gredos, 1962. - (1994). Gramatica de la llengua catalana. Descriptiva, normativa, diatbpica i diastratica. Barcelona: Enciclopbdia Catalana. Badia i Cardús, T. (1994). Aspectes del sintagma nominal en catala des de la perspectiva de la traducció automatica. Barcelona: Publicacions de 1'Abadia de Montserrat. Barnils, P. (1921). <<Apuntament de problemes de fonktica histbrica del catalb. Butlletí de Dialectologia Catalana V-IX: 56-63. Cabré i Monné, T. (1980). Sobre les preposicions febles en catala. A-EN. Universitat Autbnoma de Barcelona: Degree thesis. Calveras, J. (1928). <<Variants de les preposicions a, en, ab en els dialectes catalans,,. Anuari de llOficina Romiinica de lingüística i literatura I: 15 1-176. Cororninas, J. (1971). Lleures i converses d'unfilbleg. Barcelona: Club editor. Camprubi, M. (1984). <<Les preposicions en els complements circumstancials de lloc i de temps a través del corpus Atlas catalin. (1375). In: Miscel lania A. M. Badia i Margarit I. Barcelona: Publicacions de 1'Abadia de Montserrat. Decesaris, J.A. (1986). c<Epenthesis in Catalan,,. In: Neidle & Ndñez-Cedeño (eds.). Stlidies in Romance Languages. Dordrecht: Foris, p. 79-9 1. Fabra, P. (1918). Gramatica catalana. Barcelona: Aqua, 198 1. - (1981). Gramatica catalana. Barcelona: Teide, 10th ed., 1956. - (1984). Conversesfilolbgiques I, II. Edició crítica a cura de J. Rafel. Barcelona: Edhasa. Ferrater, G. (1981). <<Esquelets a l'armari,,. Ferrater, J. (ed.). Gabriel Ferrater. Sobre el llenguatge. Barcelona: Quaderns Crema. Gimeno, LI. (1989). Estudi lingüístic del tortosímeridional. Universitat de Barcelona: Doctoral Dissertation. - (1990). <<La reducció vochlica: e Btona iniciab. In: Estudis de llengua i literatura II. Miscel lania J. Fuster. Barcelona: Publicacions de 1'Abadia de Montserrat. 42. On these cases, see Cabré (1980). Weak Prepositions in Tortosan Catalan CatWPL 511, 1996 65 Gulsoy, J. (1965). c<The descendant of old Catalan and Proven~al ab (with)),. Revue de linguistique romane 29: 38-59. Lloret, M.R. (1988). Gemination and Vowel Length in Oromo Morphophonology. Indiana University, Bloomington: Doctoral Dissertation. - (1990). Construint on the OCP: Evidence from Vowel Length in Oromo. Universitat Autbnoma de Barcelona: unpublished ms. It8, J. (1989). <<A Prosodic Theory of Epenthesis,,. Natural Language and Linguistic Theory 7. Kenstowicz, M. (1994). Phonology in Generative Grammar. Cambridge: Blackwell. Marvh, J. (1988). Curs superior de gramatica catalana. Barcelona: Barcino, 1968. Mascaró, J. (1983). La fonologia catalana i el cicle fonoldgic. Barcelona: Publicacions de UAB. - (1984). <<Sobre la reducció de les transformacions d'elisió,,. Estudis Gramaticals I. Barcelona: Publicacions de la UAB. - (1986). Morfologia. Barcelona: Enciclopkdia Catalana. - (1987). <<A Reduction and Spreading Theory of Voicing and other Sound Effects,,. ' Catalan Working Papers in Linguistics 4.2 (1995): 267-328. - (1989). <<On the Form of Segment Deletion and Insertion Rules>>. Probus 2: 31-61. - (1995). Externa1 allomorphy in Romance. Universitat Autbnoma de Barcelona: unpublished ms. Massip, A. (1979). El dialecte tortosí a través de I'obra de Moreira. Universitat de Barcelona: Degree thesis. - (1989). Aproximació descriptiva al parlar tortosí. Tarragona: Publicacions de la Diputació de Tarragona,. McCarthy, J. (1986). <<OCP Effects: Gemination and Antigemination),. Linguistic Inquiry 17-2: 207-263. Mier, J. Z. (1986). <<Estudi sociolingüístic de certs aspectes de la llengua catalana),. Treballs de sociolingüística catalana 6: 33-1 12. Moreno, J. C. Fundamentos de sintaxis general. Madrid: Síntesis, 1987. Morales, A.; Prieto, P. (1992). <<Processos d'epkntesi en grups clítics en tortosí>>. In: Actes del Sisk ColJoqui dlEstudis Catalans a Nord-Amkrica. Barcelona: Publicacions de 1'Abadia de Montserrat, p. 131-142. Odden, D. (1988). <<Anti-gemination and the OCP>>. Linguistic Inquiry 19: 451-475. Oliva, S. (1992). La mktrica i el ritme de la prosa. Barcelona: Quaderns Crema. Palet, M.T. (1987). <<Un problema de lingüística contrastiva: las preposiciones en español y en catalam. Revista española de lingüística 17-1: 69-84. Palmada, B. (1991). La fonologia del catalti i elsprincipis actius. Universitat Autbnoma de Barcelona: Doctoral Dissertation. - (1993). <<La construcció de les síl.labes i I'elisió vocillica del catali>>. Llengua & Literatura 5: 371-392. - (1994a). <<Sobre la semblan~a excessiva en fonologia>>. Els Marges 49: 95-106. - (1994b). La fonologia del catalh. Els principis generals i la variació. Barcelona: Publicacions de la UAB. Perera, J. (1986). <<Contribució a l'estudi de les preposicions en el Tirant 10 Blanch. Primera part>>. Llengua i Literartura 1 : 5 1109. Recasens, D. (1991). Fonktica descriptiva del catala (Assaig de caracterització de la pronúncia del vocalisme i del consonantisme del catala del segle XX). Barcelona: Institut &Estudis Catalans (Biblioteca Filolbgica, XXI). - (1993). Fonktica i Fonologia. Barcelona: Enciclopkdia Catalana. 66 CatWPL 511, 1996 Esteve Clua Sancho, P. (1994). Les preposicions en catalh Valbncia: Universitat de Valbncia. Segarra, M. (1985). Histbria de la normativa catalana. Barcelona: Enciclopbdia Catalana. Solk, J. (1977). A l'entorn de la llengua. Barcelona: Laia. Wheeler, M. (1987). ctL'estructura fonolbgica de la síi.laba i del mot en catal&)>. In: Estudis de Llengua i Literatura Catalanes, XiV (Miscel.lania A.M. Badia i Margarit, 6). Barcelona: Publicacions de 1'Abadia de Montserrat, p. 79-108. Yip, M. (1988). ttThe Obligatory contour principle and phonological mles: a loss of identityn. Linguistic Inquiry 19: 65-100.