Que-deletion: the rise and fall of a syntactic fashion

xxx Estudos dedicados a David Mackenzie Universidade de Santiago de Compostela, 2015, ISBN 978-84-xxxxx-xx-x, pp. xx-xx 'Que-deletion: the rise and fall of a syntactic fashion', in Francisco Dubert García, Gabriel Rei-Doval & Xulio Sousa (eds), En memoria de tanto miragre. Estudos dedicados ó profesor David Mackenzie. (Santiago de Compostela: Universidade de Santiago de Compostela, Servizo de Publicacións e Intercambio Científico, 2015), pp. 143-159. Que-deletion: the rise and fall of a syntactic fashion CHRISTOPHER J. POUNTAIN Queen Mary, University of London 1. Introduction he phenomenon of que-deletion in Castilian is an intriguing example of a change which is not ultimately successfully embedded in the language, and its study may therefore shed light on one of the classic questions in historical linguistics identiied by Weinreich / Labov / Herzog (1968: 102), that is, why some potential changes are realised, or actuated, while others are not. By que-deletion, I mean the omission, or apparent omission, of the que complementiser in such sentences as (1b): (1) a. RogamosV1 que disculpenV2 las molestias b. RogamosV1 que disculpenV2 las molestias1 I speak of que-deletion in such cases as (1b) and regard the presence of que as the default option because there appears to be no environment in which the presence of que cannot be an alternative to its absence, and because the absence of que always appears to have been overall a minority construction. Historical accounts customarily move forward in time, but in this case I will start with the present day and work backwards, since this has the advantage of going from the known and relatively fully described to the unknown and hitherto scarcely documented. 1 In all the examples given, I indicate a deleted / omitted que by double strike-through. V1 is the main clause verb and V2 the complement clause verb. xxx 68 xxx Estudos dedicados a David Mackenzie 2. Modern Spanish here is a general view, conveniently summarised in NGLE (3230), that in modern Spanish que-deletion is (a) associated with certain formal written registers, (b) most common with the subjunctive complements of verbs of command and inluence (1b), though also found with the indicative complements of some verbs of thinking (2), (c) subject to surface contextual constraints, notably (i) resistance to material other than clitic pronouns or no intervening between V1 and V2 (compare 3a–b), and (ii) the requirement that V1 is itself in a subordinate clause (compare 4a–b). (2) Es una profesión que creoV1 que puedeV2 compararse con la suya (La Vanguardia, cit Subirats-Rüggeberg 1987: 170) (Note that this environment, where the verb and its complement constitute a relative clause, the antecedent of which is the subject of the complement verb, is one in which that-deletion is actually obligatory in English: compare It is a profession which I think (*that) can be compared with his. It should be insisted that (2) is in fact a case of que-deletion and not simply a parenthetical use of creo; (2) can be uttered, or rather, read, with no intonational junctures surrounding creo, and no commas surround creo in its written representation. Que-deletion is actually favoured where, as in this example, the construction is itself embedded in a relative clause (Keniston, 1937b: 272), when the absence of a second que is sometimes thought of as due to considerations of ‘euphony’ (Benot, 1991 [1910]: 355–7).) (3) a. Se aseguraV1 que desembarcaráV2 mañana el presidente but b. *Se aseguraV1 que el presidente desembarcaráV2 mañana (Benot, 1991 [1910]: 355–7; see also Delbecque / Lamiroy, 1999: 2026) (4) a. Un asunto que consideroV1 que tendríamosV2 que tratar ahora b. *ConsideroV1 que tendríamosV2 que tratar ahora ese asunto Another apparent context of que-deletion in modern Spanish, which we shall have further occasion to discuss in 4, is that of the complements of verbs of fearing. Here, however, we have to do not with que-deletion as such so much as with the substitution of que by no, often described (Butt & Benjamin, 2000: 330) as ‘redundant’ or ‘pleonastic’ because no does not actually negate the complement verb. However, unlike the cases of que-deletion we have so far examined, (5a), in CHRISTOPHER J. POUNTAIN Que-deletion: the rise and fall of a syntactic fashion 69 which both the que complementiser and no are present, is diferent in meaning from (5b), in which que does not appear, which is synonymous with (5c); furthermore, (5d), in which neither que nor no is present, is unacceptable. he no in (5b) is therefore most appropriately viewed as being itself a complementiser in its own right rather than simply a negative. (5) a. TemoV1 que no vengaV2 Pepe ‘I’m afraid Pepe won’t come’ b. TemoV1 que no vengaV2 Pepe ‘I’m afraid Pepe will come’ or c. TemoV1 que vengaV2 Pepe ‘I’m afraid Pepe will come’ but d. *TemoV1 vengaV2 Pepe ‘I’m afraid Pepe will come’ (Sánchez López, 1999: 2628–9) he extent and frequency of que-deletion in modern spoken Spanish is unclear. Despite the fact that Subirats-Rüggeberg (1987: 168) claims that it is ‘widespread’ (no statistics are given), it is not mentioned as a feature of colloquial Spanish by either Steel (1976) or Kany (1951); Maldonado González (1999), on the other hand, characterises it as belonging exclusively to the written language. On the basis of an admittedly very limited preliminary statistical survey, I am inclined to consider both Subirats’s and Maldonado’s positions extreme: in the Madrid habla culta corpus of 134,452 words (Esgueva / Cantarero, 1981), I have counted only two examples of creo with a que-deleted complement as opposed to 434 with a que complement, which suggests that while it is not absent from the spoken register it is extremely uncommon; in a similar search in the PRESEEA corpus based on material from Spain, 257 cases of a que complement were returned as against 3 clear cases (excluding hesitations and occurrence with intonational junctures which suggest parenthetical use of creo) of a que-deleted complement. Que-deletion is puristically castigated in modern Spanish only to the extent of not being a preferred usage, the traditional academic view being that use of que makes syntactic structure clearer (Sarmiento, 1984, 336–7; Esbozo, 517; DPD, §2.1.2, and see 5 below); this may inhibit its use in what we may regard as the standard educated language (the norma culta). he consistently indulgent prescriptive attitude towards que-deletion (because it is a cultured rather than a 70 xxx Estudos dedicados a David Mackenzie popular variant) and the fact that the variable is register- and context-dependent rather than being associated with age, sex, class or style, means that speakers themselves do not take up an attitude towards it, so allowing and even encouraging a situation of reasonably stable variation in which no immediate change is due (Silva-Corvalán, 1988: 159). 3. Que-deletion in 16th-century Spanish 3.1. Textual evidence I now contrast with the modern situation that of 16th-century Spanish, where que-deletion is certainly encountered more frequently than subsequently (Esbozo, 517; Delbecque / Lamiroy, 1999: 2026 note), though this impression has never to my knowledge ever been rigorously quantiied. What is immediately apparent is (a) that que-deletion occurred in a much more extended range of contexts, and (b) that some authors are more prone to it than others. I have analysed two nearly contemporaneous texts of rather diferent genres in which, impressionistically, que-deletion seems to achieve its heyday: Santa Teresa’s (1515-82) spontaneously written spiritual journal Libro de la vida (1562) and the irst two books of Tomás de Mercado’s (?-1575) Suma de tratos y contratos (1571), a treatise on business ethics. he list of verbs and verbal expressions for which que-deletion either in object or subject complements is attested in these texts is impressively long and is given in Table A, together with the mood (ind[icative] or subj[unctive]) of the complement verb and any other relevant syntactic information: LIBRO DE LA VIDA SUMA DE TRATOS Y CONTRATOS VERBS AND OTHER EXPRESSIONS OF ORDERING AND INFLUENCE buscar cometer ser cómodo compeler consentir convenir decir desear encargar enseñar (lo primero que enseña es...) hacer ser impedimento  subj  subj  subj  subj  subj  subj  subj  subj  subj  subj  subj subj  subj  subj CHRISTOPHER J. POUNTAIN Que-deletion: the rise and fall of a syntactic fashion impedir incitar mandar ser lo mejor ser menester ser (lo) necesario ordenar pedir  subj permitir persuadir  subj pretender procurar querer requerir rogar suplicar ser útil vedar era voluntad divina  subj  subj  subj  subj  subj  subj  subj  subj VERBS OF THINKING AND SAYING ser cierto tener por cierto creer decir entender caer en entendimiento dudar imaginar parecer pensar saber estar seguro ver temer traer temor quedar admirado hacerle al caso espantarse gustar holgarse ser gran lástima pesar placer (plega / pluguiera) ser justo ser servido tener por menos mal  subj  ind (also with no creer)  ind  ind  ind  ind  ind (also with no pensar)  ind  ind  subj, pleonastic no  subj  subj  subj  subj  subj  subj  subj  subj  subj  subj  subj  subj  subj  subj  subj  subj, pleonastic no  subj  ind  ind (also with no creer)  ind  ind  subj (negated)  ind  ind  ind (also with no pensar)  ind  ind VERBS OF FEARING  subj, ‘pleonastic’ no VERBS OF ‘EMOTION’  subj  subj (V1 negated)  subj  subj, ‘pleonastic’ no  ind  subj  subj  subj  subj  subj  subj  subj  subj  subj  subj 71 72 xxx Estudos dedicados a David Mackenzie OTHER aguardar bastar concertar dar ‘admit’ esperar no era mucho ‘it was not diicult’ poder ser restar ser (conforme a) razón  subj  subj CONJUNCTIONS caso por condición de cualquier manera dado por + adj (concessive) sino  subj  subj  ind, subj  subj  ind, subj  subj  subj  subj  subj  subj  subj  subj  subj  ind TABLE A: VERBS AND OTHER ELEMENTS EXHIBITING QUE-DELETION IN SANTA TERESA AND TOMÁS DE MERCADO It can be seen that, while the majority of items are verbs of ordering with subjunctive complements (the class which still exhibits que-deletion to the greatest extent in modern Spanish), verbs of thinking and saying with indicative complements are also strongly evidenced. In Suma, que-deletion even extends to some conjunctions (6): (6) Toda esta doctrina católica es tan verdadera que, dado que nos la enseñen estos santos doctores, los mismos gentiles autores la enseñan más largamente [...] (Suma, I.i) he main constraint on que-deletion proposed for modern Spanish, that there should be no intervening material between V1 and V2, appears not to hold for the 16th century (7), even though the V1 V2 pattern (8) is preponderant: (7) [...] que pareceV1 en queriendo comenzar a tener oración que hallamosV2 con quién hablar [...] (Vida, 27.4) (8) [...] porque días había que deseabaV1 que fueraV2 posible a mi estado andar pidiendo por amor de Dios y no tener casa ni otra cosa (Vida, 35.2) An even more striking feature of Vida is that que-deletion actually appears to be the more frequent variant with some verbs (creer, parecer, suplicar and temer: see Table C). he only item for which I have been able to establish a similar CHRISTOPHER J. POUNTAIN Que-deletion: the rise and fall of a syntactic fashion 73 preference in Suma, which has a wider range of que-deleting contexts though not proportionally so many tokens of complementation, is the conjunction dado~dado que (see (6)), which shows an overwhelming 226 cases of deletion as against 54 of non-deletion (this conjunction is not used in Vida at all). 3.2. Register, text type, style Given the strong correlation between que-deletion and register in modern Spanish, it would be logical to try and establish such data for the 16th century too. his, however, is a daunting task because of the relative infrequency of the construction and the diiculty of identifying examples by automated search procedures. I have obtained some preliminary data from CdE by the strategy of searching for patterns of rogar and rogar que followed directly by a clitic pronoun (rogar is a common verb known to favour que-deletion in the 16th century: see Table D). his retrieves a relatively large number of sequences in which rogar is followed by a complement verb without an intervening que (9a) and with que (9b) in a similar context. (9) a. rogar: V1 V2 … te ruegoV1 que me digasV2 la dispusiçión del lugar. (CdE: Cristóbal de Villalón, El Crotalón, 1552–3) b. rogar: V1 que V2 … más vos ruegoV1 que me digadesV2 si está el infante mal indignado contra mí… (CdE: Platir, 1533) For this more manageable dataset, the source of each example was then classiied according to genre (text-type); the cumulative results are given in Table B: GENRE QUE Novels of chivalry 98 QUE AS % OF TOTAL QUE 9 8.26% Other novels 27 6 18.18% Picaresque novels 6 3 33.33% Pastoral novels 6 Commercial documents 1 Chronicles 71 Travelogues 15 Religious works 137 3 21 3 67 1 13.13% 75% 24 87 33.33% 25.26% 28 42 6.25% 24.35% 38.53% 74 xxx Estudos dedicados a David Mackenzie GENRE QUE QUE AS % OF TOTAL QUE Dialogues 13 11 45.83% Drama 11 2 15.38% Letters 23 5 17.86% Mythology 1 3 75% Poetry 6 1 14.29% Proverbs 4 6 60% Overall 349 119 25.43% Table B: QUE-deletion with ROGAR + (QUE) + clitic pronoun + verb sequences in CdE What emerges is that que-deleted examples are overall in a minority, as is to be expected (119: 349, or, with que-deletion expressed as a percentage of the total, 25.43%). he genres showing a signiicantly higher percentage of quedeletion are religious works and dialogues; those showing a signiicantly lower percentage are novels of chivalry and, possibly, drama. hese preliminary results indicate that a relation between que-deletion and text-type is likely and possible to trace, and encourage further investigation along these lines. In Table C I illustrate a diferent investigative strategy, that of comparing igures for que-deletion with certain verbs in a purpose-built corpus consisting of Vida, Suma and three other roughly contemporaneous texts: the picaresque novel Lazarillo de Tormes (1554), Juan de Valdés’s Diálogo de la lengua (1535) and the irst Book of Antonio de Guevara’s Epístolas familiares (1526). As in Table B, in each cell the raw occurrences of que-deletion and non-deletion are given, followed by the expression of this ratio as the percentage of que-deletion of the total; in addition, the inal igure shows the frequency of both que-deleted and non-deleted instances in the whole text, expressed per thousand words (‰). It can be seen that there are very marked statistical diferences among these texts, in terms not only of the proportions of que-deleted and non-deleted instances, but also in the frequency of the verbs concerned, both cumulatively and individually. It can probably be concluded that Teresa is conirmed as the most proliic que deleter, while Guevara is the least proliic, and that hence not only text-type but also author are likely to prove important variables. 75 CHRISTOPHER J. POUNTAIN Que-deletion: the rise and fall of a syntactic fashion QUE/QUE QUE AS % OF TOTAL CREER PARECER QUERER SUPLICAR TEMER 1/0 100% 0.01‰ 40/6 86.96% 0.41‰ 8/6 57.14% 0.05‰ 5/2 71.43% 0.38‰ 1/0 100% 0.00‰ 2/1 66.67% 0.04‰ 12/4 75% 0.14‰ 0/2 0% 0.00‰ 2/0 100% 0.11‰ 1/0 100% 0.00‰ TOTAL TOTAL AS ‰ OF WORD-COUNT Suma 75,019 words 11/11 50% 0.29‰ 74/46 61.67% 1.06‰ 2/123 1.6% 0.42‰ 0/5 0% 0.27‰ 2/40 4.76% 1.09‰ Vida 112,868 words Guevara 294,268 words Lazarillo 18,475 words Valdés 38,473 words 3/21 12.5% 0.32‰ 376/156 70.68% 4.71‰ 1/19 5% 0.06‰ 2/9 18.18% 0.60‰ 6/31 16.22% 0.96‰ 12/16 42.86% 0.37‰ 61/54 53.04% 1.02‰ 3/127 2.31% 0.44‰ 1/8 11.11% 0.49‰ 3/39 7.14% 1.09‰ 29/49 37.18% 1.04‰ 563/327 63.25% 7.89‰ 14/277 4.81% 0.99‰ 10/24 29.41% 1.84‰ 13/110 10.57% 3.20‰ TABLE C: QUE-DELETION FOR SELECTED VERBS IN FIVE 16TH-CENTURY TEXTS 4. Que-deletion prior to the 16th century: its origins and rise Prior to the 16th century, que-deletion is too infrequent to make the study of individual texts productive, and I have therefore used the strategy, already outlined, of searching for instances of a number of verbs thought likely to be prone to que-deletion in the light of their subsequent histories, but this time exploiting CdE as a whole. Table D shows the results obtained for nine verbs in the 13th–15th centuries, the limitations of the data being indicated in the note to each verb: EXAMPLES OF QUE creer1 dudar2 guardar no(n)3 parecer2 pedir4 querer5 rogar2 suplicar6 temer no(n)7 1200S 1300S 0 0 4 0 0 0 0 0 0 2 0 12 (1) 0 2 (1) 0 0 1400S 69 6 34 27 1 12 38 133 3 76 xxx Estudos dedicados a David Mackenzie 1 Based on a full survey of creo and the pattern cre* followed by a subjunctive verb. Based on a full survey of all occurrences of the pattern du(b)d*. 3 Based only on occurrences with no(n); the pattern guard* followed by subjunctive verb returned no tokens. 4 Based on a search for the patterns pid* and ped* followed by subjunctive verb. 5 Based on a survey of quiero and the patterns quier*, quer* and quis* followed by a subjunctive verb. 6 Based on a full survey of all occurrences of the pattern sup(p)lic* and sup(p)liqu*. Sup(p)licar is rare before the 15th century. 7 Based only on ocurrences with no(n); the pattern tem* followed by subjunctive verb returned no tokens. 2 Table D: QUE-deletion, 13th–15th centuries (data from CdE) It can be seen that que-deletion is practically non-existent before the 15th century, with the striking exception of guardar no(n), which merits further investigation as a possible model for subsequent developments. Guardar has complementation patterns with both no(n) and que no(n) which are apparently identical in meaning, the same negative polarity of no(n) holding in both constructions (i.e. guárdate no(n) + subjunctive and guárdate que no(n) + subjunctive are synonymous, as shown by the English glosses in (10a-b)): (10) a. GuardateV1 que non digasV2 njnguna cosa errada contra Jacob (13th century: Alfonso X, General estoria) ‘Take care that you do not say anything false against Jacob’ b. guardateV1 que non pequesV2 con muger virgen (13th century: Sancho IV, Castigos y documentos para bien vivir) ‘Take care that you do not sin with a woman who is a virgin’ As we have seen, this is unlike the behaviour of temer in MSp, where no with que-deletion (5b) indicates positive negative-polarity and que no (5a) indicates negative negative-polarity. he same appears to be true of temer in 16thcentury Spanish too: (11) a. TemíaV1 que no habíaV2 de haber con quién me confesar [...] (Vida, 28.14) ‘I feared there would not be anyone I could confess to’ b. [...] me tenían mucho amor y temíanV1 que no fueseV2 engañada (Vida, 25.14) ‘hey loved me greatly and feared (that) I was deceived’ Despite the change of lexical item, guardar no(n) could be seen as the semantic inheritor of the function of of Latin cavēre, for which the complementiser was nē, and the variation observable in Old Castilian between guardar no(n) CHRISTOPHER J. POUNTAIN Que-deletion: the rise and fall of a syntactic fashion 77 and guardar que no(n) may represent conlicting pressure from the exemplary force of Latin cavēre nē versus the overwhelmingly general rule that verbal complements are introduced by que. However, not all the heirs of verbs which took nē as a complementiser in Latin (verbs of preventing, forbidding and fearing) initially follow the model of guardar. Embargar, the most frequent verb of preventing in Old Castilian, with a number of attestations from the 13th century onwards, appears not to undergo que-deletion, although there are a number of examples of a ‘pleonastic’ no(n) being used after the que complementiser with positive negative-polarity: (12) Ca estos siempre punnan de los embargarV1 que se no saluenV2 (CdE: Siete partidas, 13th cent.). ‘For these people always strive to prevent them from being saved’ (not: ‘prevent them from not being saved’) he verb impedir, of which there are no convincing examples before the 15th century, shows just one or two examples of que + ‘pleonastic’ no complementation in that century (13); and by the late 16th century there are examples of impedir with que-deletion both with and without no (14–15), as well as with plain que (16): (13) y otrosy en el capitulo dezeno ay veynte auctoresy veynte y siete auctoridades que aprueuan el vino blanco ser prouechoso para impidirV1 que la piedra no se engendreV2 (CdE: Julián Gutiérrez de Toledo, Cura de la piedra, 15th cent.) ‘and also in the tenth chapter there are twenty authors and twenty-seven authorities who approve white wine as being beneicial for preventing the stone from forming’ (not ‘preventing the stone from not forming’) (14) o impedirleV1 que no consigaV2 lo que tan honestamente apetece (Tomás de Mercado, Summa de tratos y contratos, 16th cent.) ‘or prevent him from obtaining what he so honestly desires’ (not ‘prevent him from not obtaining’) (15) [...] para que, sacando todos, haya abundancia y se impidaV1 que crezcaV2 el precio (ibid.) ‘so that, if they all take it out, there will be plenty and the price will be prevented from rising’ (16) y no quiere que los dioses puedan impedirV1 que crezcanV2 (CdE: Lope de Vega, La bella Aurora, end of 16th cent.) ‘and does not want the gods to be able to prevent them growing’ 78 xxx Estudos dedicados a David Mackenzie Vedar is attested only with a que complementiser before the 16th century. I have found just one example of prohibir with que-deletion and a (‘pleonastic’) no in the 15th century (17); a ‘pleonastic’ no is also possible with que (18): (17) …el qual mucho tiempo cercaron los ijos de israel porque les fue de dios prohibidoV1 que no le tocassenV2 (CdE: Bernardo de Breidenbach, tr.Martín Martínez de Ampiés, Peregrinatio in Terram Sanctam; Viaje siquier peregrinación de la tierra, 15th cent.) ‘which the children of Israel camped around for a long time because it was forbidden to them by God to make contact with it’ (not ‘forbidden not to make contact’) (18) quando prohibeV1 que no salgaV2 la orina (CdE: Julián Gutiérrez de Toledo, Cura de la piedra, 15th cent.). ‘when it stops the urine coming out’ (not ‘stops the urine not coming out’) he irst examples of the use of temer with que-deletion, accompanied by a ‘pleonastic’ no, are from the late 15th-century Celestina (19): (19) ...temoV1 que no la ayan levadoV2… (CdE: Fernando de Rojas, Comedia de Calisto y Melibea, 15th cent.). ‘I fear they have taken it’ In summary, the guardar no construction, with its apparent que-deletion, is unusual in type in medieval Castilian. It is possible that it provided the pattern for the use of no as a complementiser without que for some other verbs of related semantic classes, beginning with temer and the ‘learned’ verb prohibir (the latter again perhaps in imitation of Latin prohibēre, for which the only complementiser classically was nē). hus Prohibió (que) no saliese parallels Latin prohibuit ne exiret (in the 16th century the modern Prohibió que saliese in the same sense is still a minority construction). But it seems unlikely (although more quantitative research on 15th century and early 16th century texts is needed to establish this deinitively) that the use of no on its own as a complementiser, in imitation of Latin nē, provided a model for the suppression of que with other subjunctive-requiring verbs of forbidding and that que-deletion then difused more generally to verbs of ordering, other verbs with subjunctive complements and verbs of saying and thinking with indicative complements, since the attestation of que-deletion with the verbs of saying and thinking creer and parecer is already substantial in the 15th century and antedates that of a number of verbs more similar semantically to guardar. CHRISTOPHER J. POUNTAIN Que-deletion: the rise and fall of a syntactic fashion 79 Another causal hypothesis for the acceleration of que-deletion in the 15th century is contact inluence, Latin and Italian both being plausible candidates as such a contact source. With regard to Latin, deletion of the complementiser ut can be observed in the classical language with subjunctive-requiring verbs of command and inluence such as rogō, moneō, suādeō, imperō, cūrō, oportet, necesse est, licet and, as we have seen, caveō, a phenomenon supericially very similar to that observable in Castilian. However, the problem is that the 15th century shows a much greater range of que-deleting verbs, as demonstrated in Table D, so there is no isomorphism between Old Castilian and Classical Latin in this respect, even setting aside the issue of lexical replacements. As regards contact with Italian, the Old Castilian data likewise suggest that quedeletion in Castilian is unlike che-deletion in Italian, either in type or in chronology. Che-deletion in Old Italian ranges over a much larger number of syntactic contexts: it is attested not only with verbal complements, but also with complex conjunctions, comparative structures and restrictive object and subject relative clauses (Wanner, 1981; Poletto / Cocchi, 2007). Although the signiicant expansion of complement que-deletion in 16th-century Spanish might plausibly have been the result of the example of Italian che-deletion, which was very frequent by the 15th century, such putative inluence did not extend to the omission of que with comparatives and relative clauses in Spanish, neither of which to my knowledge is attested at all. Given the inconclusive nature of such hypotheses as the above, it is possible that the motivation for que-deletion in Spanish was of a more general structural kind. Que was and continues to be the most commonly occurring word in Spanish, marking clausal subordination in complement and relative structures, involved in comparative constructions of inequality, and also serving, especially in the spoken language, as an introductory clause marker. Some pruning of this heavy functional load could be seen as a quite natural development, and in the area of complementation could be expected to occur in contexts where the identiication of main verb and subordinate verb was clear through juxtaposition and/or diference in mood (this is consistent with Valdés’s view (see 5 below) that the complementiser que is ‘superluous’). In particular, the avoidance of two instances of que in close proximity where a complement is embedded in a relative clause may be stylistically more ‘euphonious’ (see (2) above). he same factors pertain to Italian, where, as we have seen, che-deletion was further advanced. 80 xxx Estudos dedicados a David Mackenzie 5. Demise Why did que-deletion eventually recede? he irst factor may be that it was in all probability an élitist usage which did not successfully embed in lower echelons of society and in everyday usage. Keniston (1937a: 676), in one of his characteristic tantalisingly insightful thumbnail sketches, says that Santa Teresa, in using quedeletion so prodigally, cannot be relecting popular usage, since que-deletion is rare in Lope de Rueda, hence supporting the hypothesis that it is an essentially cultured phenomenon. (We must beware, however, of assuming that Rueda only relects popular usage, since many of his upper-class characters use the high-style retórica. Yet what is interesting in this connection is that while Rueda characterises retórica with a number of its stereotypical syntactic and lexical features — verb-last order, preposed adjectives, absolute constructions — he indeed does not seem to exploit que-deletion for this purpose, which would suggest that quedeletion was not archetypically associated with retórica but was rather a matter of personal preference, or fashion; it may also be that que-deletion was more of a written than a spoken phenomenon.) Another factor in the recession of que-deletion may be the increasing favouring of transparency over economy and euphony. Prescriptive views in the 16th century cannot of course be of the same order as the Academic pronouncements of the 18th century and later, but we do have at our disposal two interestingly contrasting observations. he irst, prioritising economy, is Juan de Valdés’s (1535) view that the use of que is ‘superluous’: VALDÉS. Diríale primeramente que guardasse lo que al principio dixe de los artículos, porque esto pertenece assí para el hablar bien como para el escrivir. Avisaríale más que no curasse de un que superluo que muchos ponen tan continamente, que me obligaría quitar de algunas escrituras, de una hoja, media dozena de quees superluos. MARCIO. Dadnos algunos exemplos para que entendamos esso. VALDÉS. De refrán no se me ofrece ninguno que tenga este que demasiado, y creo lo causa la brevidad con que stán escritos, pero, si miráis en lo que leéis, hallaréis ser verdad lo que os digo en partes semejantes que ésta: creo que será bien hazer esto, adonde aquel que stá superluo, porque diría mejor: creo será bien hazer esto. [my underlining] (Lope Blanch, 1969: 154) It is diicult to know to what extent Valdés is relecting the Toledan usage he ostensibly admired and how much is a matter of idiosyncratic personal preference; what is certain is that he did not apply the principle consistently himself CHRISTOPHER J. POUNTAIN Que-deletion: the rise and fall of a syntactic fashion 81 (see Table C above which demonstrates quite clearly that in Valdés’s own writing que-deletion is a minority construction). But what this opinion probably shows that que-deletion was held in some regard in the irst half of the century. Nearly a hundred years later, however, in 1625, we ind Gonzalo de Correas soundly castigating the omission of que and exalting its presence as lending clarity to Spanish (hence prioritising transparency), giving, so he considers, Spanish an advantage in this respect over Latin: Algunos quitan la que en muchas ocasiones, ó caiendo ello ansi sin rreparar, ó por que se enfadan de ir á lo claro, i llano, i quieren buscar sainete i modo nuevo de hablar; pero sepan que se engañan los que ansi lo hazen de industria, i que dexan la rrazon manca i confusa, i que con todo se deve suplir i entender la que, i en esto de poderse quitar, i suplirse se conozerá tanbien que es partezilla... Da esta partezilla que tanta grazia i claridad á la orazion que con ella corre descansada i rredonda, i queda el animo satisfecho i quieto. Tiene mas, que rreduze las oraziones de ininitivo á los tienpos i modos initos de indicativo i subiuntivo, i en esta que por ello tiene doblada claridad y fazilidad la lengua Castellana mas que la Latina. [my underlining] (Alarcos García, 1954: 174) Correas’s view may perhaps be seen as the beginning of the Academic preference for the use of que; the reference to Latin may indicate that que-deletion was associated with formal written Latinate prose. It may also indicate a change of attitude towards the construction which arrested its development even in educated writing. In other words, que-deletion becomes unfashionable. Wanner (1981) similarly suggests that the demise of Italian che-deletion was the result of a change in taste of which Bembo was typical. 6. Conclusions Que-deletion is a syntactic fashion which has its heyday in the 16th century. Its origins may lie in contact with Latin or Italian, or may relect an evolutionary tendency towards economy, but there is at present insuicient evidence to judge the relative strength of these causal hypotheses. Its demise is most likely the consequence of its never having become suiciently irmly embedded socially, and of a purist insistence on clarity and transparency. It continues today in some registers of Spanish, but has only the status of a stylistic indicator, and coexists with presence of the complementiser in a situation of stable variation. 82 xxx Estudos dedicados a David Mackenzie References Alarcos García, Emilio (1954): Gonzalo Correas: Arte de la lengua española castellana. Madrid: CSIC. Benot, Eduardo (1991 [1910]): Arte de hablar. Gramática ilosóica de la lengua castellana (Reproducción facsímil). Barcelona: Antropos. Bosque, Ignacio / Violeta Demonte (eds.) (1999): Gramática descriptiva de la lengua española, 3 vols. Madrid: Espasa. Butt, John / Carmen Benjamin (2000): A New Reference Grammar of Modern Spanish, 3rd ed. London: Arnold. Delbecque, Nicole / Béatrice Lamiroy (1999): ‘La subordinación sustantiva: las subordinadas en los complementos verbales’, in Bosque / Demonte (eds.), ii, 1965–2081. Esbozo = Real Academia Española (1973): Esbozo de una nueva gramática de la lengua española. Madrid: Espasa-Calpe. Esgueva, M. / M. Cantarero, 1981: El habla de la ciudad de Madrid. Madrid: CSIC. Kany, Charles E. (1951): American-Spanish Syntax, 2nd ed. Chicago: University of Chicago Press. Keniston, Hayward M. (1937a): The Syntax of Castilian Prose. Chicago: University Press. Keniston, Hayward M. (1937b): Spanish Syntax List. New York: Holt. Lope Blanch, Juan M. (1969): Juan de Valdés: Diálogo de la lengua. Madrid: Castalia. Maldonado González, Concepción (1999): ‘Discurso directo y discurso indirecto’, in Bosque / Demonte (eds.), ii, 3549–3595. NGLE = Real Academia Española / Asociación de Academias de la Lengua Española (2009): Nueva gramática de la lengua española. Morfología / Sintaxis I. Madrid: Espasa Libros. Poletto, Cecilia / Gloria Cocchi (2007): ‘Complementizers: zero, one or two?’, CIDSM - 2nd Cambridge Italian Dialect Syntax Meeting, University of Cambridge, January 26th-27th 2007 [http://www.mml.cam.ac.uk/italian/news/CIDSM2web/ Abstracts%20pdf ’s/Poletto-Cocchi.pdf ]. Sánchez López, Cristina (1999): ‘La negación’, in Bosque / Demonte (eds.), II, 2561– 2634. Sarmiento, Ramón. (ed.) (1984): Gramática de la lengua castellana 1771, Edición facsímil y apéndice documental. Madrid: Editora Nacional. Silva-Corvalán, Carmen (1988): Sociolingüística: teoría y análisis. Madrid: Alhambra. Steel, Brian (1976): A Manual of Colloquial Spanish. Madrid: SGEL. CHRISTOPHER J. POUNTAIN Que-deletion: the rise and fall of a syntactic fashion 83 Subirats-Rüggeberg, Carlos (1987): Sentential Complementation in Spanish. Amsterdam/Philadelphia: Benjamins. Wanner, Dieter (1981): ‘Surface complementizer deletion: Italian che ~ Ø’, Journal of Italian Linguistics, 6, 48–82. Weinreich, Uriel / William Labov / Marvin I. Herzog (1968): ‘Empirical foundations for a theory of language change’, in W.P. Lehmann / Yakov Malkiel (eds.), Directions for Historical Linguistics (Austin: University of Texas Press), 95-189. Texts ans corpora CdE = Davies, Mark (2002-): Corpus del Español: 100 million words, 1200s-1900s. [http://www.corpusdelespanol.org]. Consultado [28 de setiembre de 2014]. Guevara = Guevara, Antonio de (1526): Libro primero de las epístolas familiares, edición digital basada en la edición de Madrid, Aldus, 1950-1952. [http://www.cervantesvirtual.com/obra/libro-primero-de-las-epistolas-familiares--2]. Lazarillo = Anon. (1554): Lazarillo de Tormes, Edición de Francisco Rico (1987), 2nd ed. Madrid: Cátedra. Also [http://www.cervantesvirtual.com/obra-visor/la-vida-delazarillo-de-tormes-y-de-sus-fortunas-y-adversidades--0/html/fedb2f54-82b111df-acc7-002185ce6064_2.html#I_0_]. PRESEEA (2014-): Corpus del Proyecto para el estudio sociolingüístico del español de España y de América. Alcalá de Henares: Universidad de Alcalá. [http://preseea. linguas.net]. Consultado: [30 de marzo de 2015]. Suma = Mercado, Tomás de (1571): Suma de tratos y contratos, edición digital basada en la edición de Madrid, Instituto de Estudios Fiscales, Ministerio de Economía y Hacienda, 1977 [http://www.cervantesvirtual.com/obra/suma-de-tratos-y-contratos--0], Parts I and II. Teresa = Teresa de Jesús, Santa (1562): Libro de la vida, Edición de Dámaso Chicharro (1994), 10th ed. Madrid: Cátedra. Also [http://www.cervantesvirtual.com/ obra/la-vida-de-la-santa-madre-teresa-de-jesus-y-algunas-de-las-mercedes-quedios-le-hizo-escritas-por-ella-misma--0]. Valdés = Juan de Valdés (1535): Diálogo de la lengua, Edición de Juan M. Lope Blanch (1969). Madrid: Castali d a.