From: Cervantes: Bulletin of the Cervantes Society of America 10.2 (1990): 3-14.
Copyright © 1990, The Cervantes Society of America

ARTICLE

DANIEL EISENBERG

	Los que son pusilánimes, descuidados y de pecho flaco suelen no pronunciar la h en las dicciones aspiradas como eno por heno y umo por humo, etc.
	Sebastián de Covarrubias2

number of Spanish consonants changed between medieval and modern Spanish. These changes took place irregularly, and each has its own chronology and geography. In several cases, therefore, different pronunciations coexisted in Golden Age Spain.³ The

     ¹ I would like to thank Máximo Torreblanca, Ralph Penny, Douglas Gifford, and James Wyatt for their invaluable comments on drafts of this article.
     ² Tesoro de la lengua castellana o española, ed. Martín de Riquer (Barcelona: Horta, 1943), p. 672.
     ³ Contemporary discussions of Castilian phonetics, such as that included in Juan de la Cuesta's Libro y Tratado para enseñar leer y escriuir breuemente y con gran facilidad cõ reta pronunciacion y verdadera ortographia todo Romance Castellano, y de la distincion y diferencia que ay en las letras consonãtes de vna a otras en su sonido y pronunciacion (Alcalá: casa de Juan Gracián, 1589), are valuable only as primary sources. They are conservative, prescriptive rather than descriptive, and lack modern linguistic concepts. I have used as sources: Amado Alonso, De la pronunciación medieval a la moderna en español, ultimado y dispuesto para la imprenta por Rafael Lapesa (Madrid: Gredos, 1955-69); Rafael Lapesa, Historia de la lengua española, 8th edition (Madrid: [p. 4] Gredos, 1980), pp. 367-81; Emilio Alarcos Llorach, “Fonología diacrónica del español,” in his Fonología española, 4th edition (Madrid: Gredos, 1965); and Douglas Gifford, “Spain and the Spanish Language,” in Spain. A Companion to Spanish Studies, ed. P. E. Russell (London: Methuen, 1973), pp. 21-22. Valuable for background is H. Tracy Sturken, “Basque-Cantabrian Influence on Alfonsine Castilian,” Studia Neophilologica 41 (1969), 298-306, and Thomas J. Walsh surveys in detail “Spanish Historical Linguistics: Advances in the 1980s,” Hispania, 73 (1990), 177-200, treating consonants on pp. 178-79 and 191-92. A recent overview of the topic of sixteenth-century phonetic changes, on which he says “desde hace más de cuarenta años, no creo que se haya escrito en nuestro dominio tanto como de las cuestiones aludidas en el título,” is offered by Emilio Alarcos Llorach, “De nuevo sobre los cambios fonéticos del siglo XVI,” Actas del I Congreso Internacional de Historia de la Lengua Española (Madrid: Arco, 1988), pp. 47-59.

3

4	DANIEL EISENBERG	Cervantes

Cervantine editor, wishing to assess the costs of modernization, needs to know what Cervantes' pronunciation was. In short, our topic is the cases in which the links between the spellings of the principes editions and Cervantes' sounds are unclear. Little attention has been paid to this question.⁴
      One might think that Cervantes' autographs, free of possible compositorial distortion, would offer material for a study of his

     ⁴ In contrast, Shakespeare's pronunciation has been extensively studied. The most recent entrega is that of Fausto Cercignani, Shakespeare's Works and Elizabethan Pronunciation (Oxford: Clarendon Press, 1981); important predecessors are Helke Kökeritz, Shakespeare's Pronunciation (New Haven: Yale University Press, 1953), on whom Cercignani has some harsh words, and Wilhelm Viëtor, A Shakespeare Phonology (1906; rpt. New York: Ungar, 1963), on whom Kökeritz has harsh words. Curiously, for none of these scholars was English a first language.
     On Cervantes' pronunciation, the only serious discussion is Francisco Rodríguez Marín's appendix on “La x de Quixote” in his “nueva edición crítica” (Madrid: Atlas, 1947-49), IX, 20-32. Joaquín López Barrera, in Cervantes y su época (Madrid, 1916), a book not for cervantistas according to its own introduction, presents on pp. 143-46 guidelines on how to pronounce the “suave” language of Cervantes. These are based on general notions of Golden Age pronunciation (s pronounced differently from ss, for example) rather than study of Cervantes. Another elementary discussion is found in Juan B. Selva, “La gramática y el Quijote,” Boletín de la Academia Argentina de Letras, 16 (1947), 641-49, at pp. 645-48. I have not seen the articles of Conrado Muiños Sáenz, “La pronunciación de la x,” El Averiguador Universal, March 31, 1881, and “¿Cómo pronunciaba el nombre de Don Quijote?,” Revista Agustiniana 7 (1884), 199-204 and 8 (1884) 489-97, both cited by Raymond Grismer, Cervantes: A Bibliography [Vol. I] (New York: H. W. Wilson, 1946), p. 110.
     Editors of Cervantes have either modernized completely and uncritically, or, nearly as uncritically, have taken fidelity to the first edition's spelling as a standard of purity and editorial virtue. The only editors to have studied questions of modernization are Allen and Schevill-Bonilla.

10.2 (1990)	Cervantes' Consonants	5

phonetics.⁵ Yet the use of the autographs is filled with practical difficulties. Miguel Romera-Navarro, author of the only monograph on them, confessed his difficulty in deciding which were authentic. Despite his caution, he failed to identify what Rodríguez-Moñino called a “fals[ificación] . . . evidente y notoria,”⁶ and thus his conclusions on Cervantes' handwriting are contaminated by a forgery. The reading of the facsimiles is difficult and confusing,⁷ and transcriptions contain errors and regularization. The documents written in Cervantes' own hand do confirm that the irregular spelling of his published works is not, or not merely, the work of compositors, but they leave unanswered the question of the sounds which the letters were intended to represent.
      Another potential source for information on Cervantes' consonants is his spelling of words from other languages. The principle is well established in historical linguistics: that “Cicero” is spelled in Greek inscriptions with kappas shows the Classical Latin pronunciation of c before e or i. Spanish missionaries' spelling of Mexican languages has been studied as evidence for their pronunciation of Spanish,⁸ and Spanish transcription of Arabic words, and the reverse, has been similarly used.⁹

     ⁵ Ten autographs are reproduced and edited by Manuel Romera-Navarro, Autógrafos cervantinos, University of Texas Hispanic Studies, 4 (Austin: University of Texas, 1954); to these should be added the letter published by Agustín G. de Amezúa, “Una carta desconocida e inédita de Cervantes,” BRAE, 34 (1954), 217-23, and from them subtracted the letter studied in the article of Rodríguez Moñino, cited in the following note.
     ⁶ Antonio Rodríguez-Moñino, “La carta de Cervantes al cardenal Sandoval y Rojas,” NRFH, 16 (1962), 81-89, at p. 85.
     ⁷ As I wrote previously, “se puede entender, si ésta es su letra, cómo un compositor leyó . . . ‘ceremonias’ en vez de las ‘cirimonias’ de Sancho” (Las semanas del jardín de Miguel de Cervantes [Salamanca: Diputación, 1988 (1989)], p. 140). On a related point, see Helena Percas de Ponseti, “A Revision: Cervantes's Writing,” Cervantes 9.2 (1989), 61-65.
     ⁸ Delos L. Canfield, Spanish Literature in Mexican Languages as a Source for the Study of Spanish Pronunciation (New York: Instituto de las Españas [Columbia University], 1936).
     ⁹ Amado Alonso, “Correspondencias arábigo-españolas en los sistemas de sibilantes,” RFH, 8 (1946), 12-76; Máximo Torreblanca, “La ‘s’ hispano-latina: el testimonio árabe,” RPh, 35 (1982), 447-63; Juan Martínez Ruiz, “Lenguas en contacto: hispanoárabe granadino y castellano de repoblación,” in Actas del I Congreso Internacional de Historia de la Lengua Española (Madrid: Arco, 1988), I, 149-63. I have not seen Arnald Steiger, Contribución a la fonética del hispanoárabe y de los arabismos en el ibero-románico y el siciliano (Madrid, 1932).

6	DANIEL EISENBERG	Cervantes

Cervantes' writing of Italian has been cited as evidence for his pronunciation of intervocalic x (Alarcos, p. 56); the spellings Quichotte and Chisciotte of the early translators have long been cited as evidence of how the translators thought the consonant was pronounced in Spanish.
      The many Arabic and Turkish words and names found in Cervantes' works provide a considerable body of additional source material. However, the Arabic found in Cervantes' works is “un árabe coloquial . . . [típico] de los dialectos árabes magrebíes,” and we find “cierto afán por parte de Cervantes de acomodar el árabe a la fonética propia del español.”¹⁰ The present writer does not feel competent to undertake the analysis of such data.
      A more manageable source is Cervantes' poetry. Rhyme and meter, combined with knowledge of word history, provide a framework with which to determine pronunciation.¹¹ For example, rhyme confirms that, as would be expected, Cervantes did not pronounce the Latinate consonants in such clusters as -ct- and -mpt-. Sonetos and tercetos never had a -ct-, so if perfectos is rhymed with them (21, 7-9-11),¹² it was pronounced perfeto. We also find trasumpto was rhymed with junto (42, 8-10), so it was pronounced trasunto.¹³ Either Cervantes wrote the more learned, “correct” spellings perfectos and trasumpto, or they are the product of his typesetters.
      The same principle can be used to study the pronunciation of individual letters. In some areas the newer voiceless intervocalic s coexisted, in Cervantes' day, with the older voiced s. (The

     ¹⁰ J. M. Sola-Solé, “El árabe y los arabismos en Cervantes,” in Estudios literarios de hispanistas norteamericanos dedicados a Helmut Hatzfeld con motivo de su 80 aniversario, ed. Josep M. Sola-Solé, Alessandro Crisafulli, and Bruno Damiani (Barcelona: Hispam, 1974), pp. 209-22, at p. 222.
     ¹¹ Manuel Alvar applies rhyme to the study of consonants in “Valor fonético de las rimas en la Gaya ciencia de Pedro Guillén de Segovia,” Anuario medieval, 1 (1989), 10-33, and José Muñoz Garrigós speaks “Sobre unas rimas anómalas con sibilante,” in Homenaje a Álvaro Galmés de Fuentes, II (Oviedo: Universidad de Oviedo and Madrid: Gredos, 1985), 131-50.
     ¹² Throughout this article, volume, page, and line references are to the only edition of Cervantes' complete works with line numbers, that of Rudolph Schevill and Adolfo Bonilla y San Martín (Madrid: the editors, 1914-41). If a title is not specified, the reference is to the Viage del Parnaso.
     I have modernized the following pairs of graphemes, of no phonetic significance: u/v, i/j, and i/y. (See Daniel Eisenberg, A Study of Don Quixote [Newark: Juan de la Cuesta, 1987], p. xxiv.)
     ¹³ Also note the rhyming of Calisto with quisto and visto (25, 27-29-31).

10.2 (1990)	Cervantes' Consonants	7

voiced s resembled the sound we know in English as z; intervocalic voiced s is common in Italian.) Which was Cervantes' usage? The words caso and casa would have had a voiced s if any did, yet they are rhymed with passo(a), in turn rhymed repeatedly with Parnaso, which must have been voiceless (13, 8-10; 16, 2-4-6; 23, 3-5-7; 106, 3-5-7). Queso is rhymed with sucesso and peso (16, 23-25-27). Cervantes' intervocalic s was thus the familiar voiceless s of modern Spanish (an apicoalveolar sibilant). Ss represented the same sound. Vase (va + se), which must have been voiceless and is rhymed with hablasse and passe (39, 15-17-19), also shows that a single s could represent the voiceless sound.
      B and v, which in writing were much closer than they are in type, represented the same sound. In the Parnaso he rhymes sabes, graves, and alabes (23, 27-29-31), suave, grave, and cabe (38, 26; 39, 2-4), nuevo, Febo, and llevo, (40, 31; 41, 1-3), aumentativa, arriba, and oliva (50, 14-16-18), etc. Cervantes always signed his name with a b, yet allowed it to always be printed on the title pages of his books with a v.¹⁴ If they were pronounced the same, the sound of the intervocal b/v was almost certainly the modern bilabial fricative .¹⁵ That they were pronounced the same suggests that the spelling of Sancho's vaziyelmo (rather than baciyelmo, as modern editors emend it), resembling vazío as much as bacín, may not have any implication at all.¹⁶
      C before e or i, ç, and z are also used interchangeably. -Za is repeatedly rhymed with -ça (l4, 30-32 and 15, 2; 55, 12-14-16; 86, 32 and 87, 2-4; 88, 10-12-14; 108, 5-7- 9), and -ços is rhymed with -zos (39, 27-29-31). Cabeça is found almost simultaneously with cabeza (91, 17). In one of his earliest published poems he rhymes engrandeze (with grandeza on the same line) with paresce and resplandesce (“Elegía al Cardenal Espinosa,” Poesías sueltas [Comedias y entremeses, VI], 17, 7-9-11). There is no confusion between these letters and s/ss; we never find, in either his published texts or his autographs, such forms as sielo or sapato, common in Andalusian writers.¹⁷ Therefore, c before e or i, ç,

     ¹⁴ Eisenberg, “On Editing Don Quixote,” Cervantes, 3 (1983), 3-34, at pp. 22-23.
     ¹⁵ See Alonso, De la pronunciación medieval a la moderna, Chapter I.
     ¹⁶ I suggested such an implication in “On Editing,” p. 11.
     ¹⁷ C before e or i, s, and z are frequently confused in the little-known texts published by Manuel Gómez-Moreno, Unos borradores cervantescos (Barcelona, 1945), and this is a strong argument against their authenticity.

8	DANIEL EISENBERG	Cervantes

and z were not pronounced with the familiar apicoalveolar sibilant s referred to above. It is unlikely that they were pronounced  (the familiar Castilian pronunciation of z), a sound which was not “extendido antes de la segunda mitad entrada del XVII” (Alarcos, p. 273). Its predecessor, and surely Cervantes' pronunciation, was a voiceless dental sibilant (the modern English s).¹⁸ G before e or i, j, and intervocalic x all had the same pronunciation. Orejas, alexas, and quexas are rhymed (48, 9-11-13), as are trafalmeja, vieja, and dexa (59, 16-18-20), viejo, sobrecejo, and perplexo (76, 21-23-25), dixo, prolixo, and hixo (76, 24-26-28), roxa, floxa, and antoja (91, 1-3-5), Tajo, trabaxo, and baxo (106, 12-14-16), and dixo, hijo, and fixo (113, 31; 114, 1-3). (I have not found any verses ending in -ge(r) or -gir, and -gi is impossible, but Rodríguez Marín points out that both gimio and ximio are found in Don Quixote [I, 152, 35 and IV, 18, 32, respectively].) All of these, if they were pronounced identically, must have been voiceless. The subsequent change of this voiceless sound to the modern jota had been made in some parts of Spain, but was far from generalized and was almost certainly not Cervantes' own pronunciation, which was the predecessor of the jota, š.¹⁹
      Was Cervantes lleísta or yeísta?²⁰ Words with intervocalic y are never rhymed with those with intervocalic ll: it is ponella, ella, bella (14, 6-8-10), halla, canalla, calla (59, 13-15-17), but suyo, arguyo, tuyo (15, 19-21-23). We must conclude that Cervantes pronounced the ll differently, and he was, therefore, lleísta.
      The situation with the h —whether Cervantes pronounced it or not— is more complex.²¹ The aspiration of h derived from

     ¹⁸ On the two pronunciations of the s in Golden Age Spanish, see D. Lincoln Canfield, “Spanish ç and s in in the Sixteenth Century: A Hiss and a Soft Whistle,” Hispania, 33 (1950), 233-36. Canfield's proposed pronunciation of the ç has been refuted by Alonso.
     ¹⁹ This is the question discussed by Rodríguez Marín (see note 4). He defends the jota as the sound with which Cervantes pronounced these letters, and quaintly characterizes the š pronunciation as “gachón” and “blanducho” (p. 30). Rodríguez Marín was answered by Américo Castro (the references are in the appendix cited), and seems himself aware that Castro's answer is unrefutable; his own examples do not support a velar phoneme. The historical linguists Lapesa and Gifford both cite the foreign spellings Quichotte and Chisciotte as evidence for a voiceless palatal x.
     ²⁰ A lleísta pronounces the ll like the Italian gli: Castiglia. A yeísta would pronounce it Castiya.
     ²¹ John Lihani calls the pronunciation of the h “uno de los problemas más enmarañados y enredados de la lingüística española . . . . En cualquier [p. 9] época que escojamos y cualquiera que sea la región de España, la historia de la h- procedente de la f- latina ha sido distinta” (El lenguaje de Lucas Fernández [Bogotá: Instituto Caro y Cuervo, 1973], p. 122). Lihani presents a helpful overview of the problem, pp. 122-49 and 169-71. The h is also discussed by R. Thomas Douglass, “The Letter H in Spanish,” Hispania, 70 (1987), 949-51, and in the works cited in note 3 and in later notes in this article.

10.2 (1990)	Cervantes' Consonants	9

Latin h, as with the word hora, had been lost in Roman times, and was not present in church Latin. The h of huevo and other words beginning with hue- was never aspirated.²² However, many sixteenth-century speakers from the southern half of Spain, as Cervantes and his parents were, aspirated the h derived from Latin f. (It is found on words such as humo and hermoso.) During Cervantes' lifetime this aspiration was disappearing. The later an author's birthdate, the less likely aspiration.²³ The center of the change was the new capital Madrid; the sound change was brought to Madrid by Felipe II's new bureaucracy, emigrated from Castilla la Vieja in an aftershock of the so-called reconquista (Lapesa, p. 372). As it affected metrics it seems to have been highly visible in literary circles, as is suggested by the passage from Covarrubias quoted at the outset.
      In short, Cervantes certainly did not aspirate the h of such words such as honor and hoy (henceforth referred to as /h/). He might have aspirated the h of words such as hermoso and hacer (henceforth referred to as /f/). Did he?
      In the first place, it is clear that Cervantes distinguished in spelling between these two types of h's. In his books, h is often omitted on words beginning with /h/, such as onor, oy, aver, Omero. In contrast with the madrileño Lope,²⁴ I have found no examples of omission of h on words beginning with /f/, such as

     ²² The u and v were first used as we do today, with u only a vowel and v a consonant, in 1726 (Lapesa, p. 422). Until that time they were “dos dibujos de una sola letra” (Alonso, De la pronunciación medieval a la moderna, I, 15). V was used at the beginning of a word and u in the middle, producing such forms as vna, vua (uva), huuo, etc. The word huevo would have been written veuo, and could be read as vevo, i.e. bebo. To avoid this potential confusion, the convention was adopted of adding an initial h to words beginning with ue: Huesca, huerto, hueso, etc.
     ²³ Eduardo Benot, Prosodia castellana y versificación (Madrid: Juan Muñoz Sánchez, n.d. [1892]), II, 392.
     ²⁴ Walter Poesse, The Internal Line-Structure of Thirty Autograph Plays of Lope de Vega (Bloomington: Indiana University, 1949), p. 62. There is an important review of Poesse's book by W. L. Fichter, HR, 18 (1950) 269-73.

10	DANIEL EISENBERG	Cervantes

hermoso.²⁵ On Flores' list of words whose varying spelling in Cervantes' works he has studied, a varying h only represents /h/.²⁶ A typesetter might have added h to words which lacked it, but would not remove it from words which had it. If a typesetter was editing while composing, correcting Cervantes' h's, he would have done so with all h's, not just those derived from /f/. Therefore, the missing initial h on many words beginning with /h/, and its presence on all words beginning with /f/, is Cervantine, though largely obscured by the typesetters' restoration of h to many words with /h/. This is in harmony with a conclusion of Flores (p. 88), who states from his analysis of compositorial spelling preferences that Cervantes wrote some forms of haber without an initial h. That compositors intervened in this way supports the hypothesis that the learned consonant clusters found in Cervantes' works were also restored by the printers.
      In Don Quixote, an initial f is used to produce pseudoantiquity only with /f/. It is found on fermosura, malferido, and forms of hacer such as fagades, but never *foy, *fierba, *fonor. Of course this shows awareness of the history of /f/ and /h/.²⁷
      In Cervantes' verse, there is frequent synalepha of words beginning with /f/.²⁸ Of the sample studied,²⁹ it is found most often in the Viage del Parnaso (73%), slightly less so in the “Canto

     ²⁵ As source, lacking an available electronic text, I have used Carlos Fernández Gómez' Vocabulario de Cervantes (Madrid: Real Academia Española, 1962), which does not modernize spelling.
     ²⁶ The Compositors of the First and Second Madrid Editions of Don Quixote Part I (London: Modern Humanities Research Association, 1975), chart one, between pp. 10 and 11. Flores studies the words hábito-ábito, huvo-uvo, hombro-ombro, ahí-aí.
     ²⁷ This device is studied by Francisco López Estrada, “La risible Fermosura, un rasgo de la comicidad inicial del Quijote,” Anthropos, 100 (1989), vi-ix.
     ²⁸ For discussion of hiatus and synalepha, and the roles of word stress and aspirated h (/f/) in them, see Benot (supra, note 23), II, 389-99; S. Griswold Morley, “Ortología de cinco comedias autógrafas de Lope de Vega,” Estudios eruditos in memoriam de Adolfo Bonilla y San Martín (1875-1926) (Madrid: Facultad de Filosofía y Letras de la Universidad Central, 1927-30), I, 525-44, especially pp. 536-39; and Poesse (supra, note 24), pp. 61-63. Some more recent bibliography and a history of the topic were provided by Homero Serís in one of his last writings, the new chapter on “Ortoepía” in his Manual de bibliografía de la literatura española, segundo fascículo de la primera parte, “1968 printing” [actually second edition] (New York: Las Américas, 1968), pp. 917-22.
     ²⁹ I have used as sources the Viage del Parnaso, all the verse in Don Quixote, and the “Canto de Calíope.”

10.2 (1990)	Cervantes' Consonants	11

de Calíope” (67%), and not at all in the “Canción desesperada” of Grisóstomo, one of the few pieces from Cervantes' works for which any manuscript exists, and whose earlier dating than Don Quixote is to my knowledge accepted. The frequency of synalepha is a tool of potential value for shedding light on the vexing question of the chronology of Cervantes' drama.³⁰

      Synalepha before /f/ is not always found. Significantly, it is sometimes lacking even when an atonic syllable begins the second word:

satisfazer al misero hambriento (51, 16)
Uno de los del numero hambriento (69, 10)
Parecia mayor su hermosura (86, 1)

Again this contrasts with the younger madrileño Lope, in whose verse such hiatus is rare.³¹ Finally, and once again in sharp contrast with Lope (Poesse, pp. 71-72), I have found no instance of hiatus before /h/ in Cervantes, not even before a tonic syllable.
     All of this suggests that Cervantes' pronunciation of /f/ was aspirated. In contrast with Lope.³²

     In conclusion, the Cervantine consonants which emerge from this analysis are an unexceptional system. They constitute good toledano, praised in the Parnaso (91, 6) and by Sancho Panza (Don Quixote, III, 244, 21-25).³³ They can be reproduced by following the above guidelines. However, in the case of h and consonant clusters, Cervantes' pronunciation is obscured by the spelling of the first editions; knowledge of word history is required as well. Reproduction of his pronunciation requires use of the unfamiliar antecessor of the modern , a dental sibilant.

     ³⁰ For introduction and references, see A Study of Don Quixote, pp. 53-54, n. 23.
     ³¹ “Hiatus [before an atonic syllable] was very unusual with Lope” (Poesse, p. 64). “Hiatus before an aspirate h and an unstressed vowel, as well as before an unstressed vowel alone, must, therefore, be considered abnormal in Lope” (Poesse, p. 77).
     ³² “H from Latin f was not pronounced [by Lope]” (Poesse, p. 62).
     ³³ Toledano was the most widely praised variety of Castilian, and Toledan usage had legal status as Spain's linguistic standard (F. González Ollé, “Un informe de 1576 sobre el habla de Toledo y su aplicación como modelo idiomático,” in Homenaje a Eugenio Asensio [Madrid: Gredos, 1988], pp. 215-23). Alcalá de Henares, Cervantes' birthplace, was lingusitically part of Toledo, which Madrid was not.

12	DANIEL EISENBERG	Cervantes

      Modernization of the spelling and pronunciation of his consonants, however, except for h and consonant clusters costs surprisingly little.³⁴ Modernizing dixo to dijo accepts Spanish's separation from Latin more than many Golden Age figures felt comfortable with, as does leaving conceto instead of the Latinate restoration concepto. Yet changing Dulzinea and Dulçinea to Dulcinea, Pança to Panza does not distort the sounds behind the spelling. And even when the sounds are changed, by pronouncing, say, the modern and jota in place of their predecessors, different sounds are assigned to two phonemes but the phonemic system remains intact.
      Of course modernization alters Cervantes' spelling and the compositors' improvements on it which, to our knowledge, he found tolerable. (While printers are criticized in Cervantes' works, there is no comment on their spelling preferences.) Yet Cervantes' spelling is perhaps less interesting to us than the sounds behind the spelling. Restoration of h to oy, Omero, and Eliodoro removes the distinction, in Cervantes' phonemic system, between these words and those beginning with /f/, such as hermoso and humo. It conceals his sounds with a veil of Latinity, as does the restoration of Latinate consonant clusters (-ct-, -mp-, etc.). It is questionable whether Cervantes, no enthusiast of Latin language and literature, would have desired this.³⁵ Both of these incomplete restorations were seemingly carried out by Cuesta and his men. One wonders whether modern editors, when modernizing Quixote and Pança, might not want to de-modernize hoy and perfecto.

     ³⁴ English spelling was and remains much more anarchic than that of Spanish, and the problems of modernization are more complex. Still, the general editor of the Oxford Shakespeare, Stanley Wells, makes a strong case for modernization in “Old and Modern Spelling,” Chapter 1 of Re-Editing Shakespeare for the Modern Reader (Oxford: Clarendon, 1984).
     ³⁵ “Todos los poetas antiguos escrivieron en la lengua que mamaron en la leche, y no fueron a buscar las estrangeras para declarar la alteza de sus conceptos. Y, siendo esto assí, razón sería se estendiesse esta costumbre por todas las naciones, y que no se desestimasse el poeta alemán porque escrive en su lengua, ni el castellano, ni aun el vizcaíno que escrive en la suya” (Don Quixote, III, 205, 27-206, 3). On Cervantes' placing Spanish authors ahead of Latin ones, see A Study of Don Quixote, pp. 75-76, and “Cervantes and Tasso Reexamined,” KRQ, 31 (1984), 305-17, at p. 306 (an updated translation of this article is about to appear in my Estudios cervantinos [Barcelona: Quaderns Crema, in press]).

10.2 (1990)	Cervantes' Consonants	13

      A final observation. Cervantes was obviously exposed, as all but the isolated were, to the phonetic diversity of Golden Age Castilian. A highly language-conscious writer,³⁶ intent on painting reality, Cervantes mentions but does not criticize this phonetic variety. The different pronunciation of gypsies is pointed out, but not censured;³⁷ when Sancho says “cirimonias” (a linguistic pincelada that a typesetter obscured), the duchess is only amused by it (Don Quixote, III, 409, 1). What Cervantes censures, rather, is the syntax of the vizcaínos,³⁸ and the garbling and misuse of learned words by the ignorant. He was semantically and lexically exacting, calling for authors to write “a la llana, con palabras significantes, honestas y bien colocadas,”³⁹ with “el lenguaje puro, el propio, el elegante y claro.”⁴⁰ Yet Cervantes was phonetically tolerant. This, I believe, gives a needed perspective to the whole question.

      In 1989 my facsimile and modernized edition of the putative Semanas del jardín fragment appeared.⁴¹ If it is indeed an Cervantine autograph, as I believe, it is the longest one known, as well as the only autograph fiction. Its value for establishing Cervantes' phonetics and spelling could be immense. As the attribution is still sub iudice, however, it could not be used as a source for this article.
      Still, it is worth examining whether the phonetics and spelling of the fragment would enhance or detract from the case for attribution. The editorial criteria followed, plus the textual

     ³⁶ Important statement of Cervantes' linguistic virtuosity, chronologically by date of first publication: Helmut Hatzfeld, El Quijote como obra de arte del lenguaje, 2nd edition (Madrid: CSIC, 1966); Leo Spitzer, “Linguistic Perspectivism in the Don Quijote,” reprinted with introduction in Spitzer's Representative Essays, ed. Alban K. Forcione, Herbert Lindenberger, and Madeline Sutherland (Stanford: Stanford University Press, 1988), pp. 222-71; Monique Joly, “Cervantes et le refus des codes: le problème du ‘sayagués,’” Imprévue (1978), 122-45.
     ³⁷ “Como gitana, hablava ceçeoso, y esto es artificio en ellas, que no naturaleza” (“La gitanilla,” Novelas exemplares, I, 41, 25-26).
     ³⁸ In Chapter 8 of Part I of Don Quixote, and in Act I of La casa de los zelos.
     ³⁹ Don Quixote, I, 37, 25-26. “Palabras claras, llanas y significantes” are requested at Don Quixote III, 245, 6-7.
     ⁴⁰ Don Quixote, III, 244, 30-31.
     ⁴¹ Note 7, above.

14	DANIEL EISENBERG	Cervantes

notes, permit one to see easily some of the ways in which the manuscript's spelling differs from modern Spanish. B-v; c before e or i, ç, and z; and g before e or i, j, and intervocalic x are changed so frequently that I included them in a list of changes made without annotation. There is no instance of an intervocalic or initial s used in place of z, nor is y used in place of ll. The popular (simpler) consonant clusters are used, and of the learned combinations, only ch is found (charidad 1:22, charater 4:13), perhaps by influence of the often-written “Christo.”⁴²
      The author of the fragment had an aspirated /f/. While the conjunction e is always substituted for y before initial (h)i, as in the modern system (6:22, 7:17, 8:14, 8:23, 9:3, 12:18, 13:12), y is used three times before the word hijo (3:1, 7:7, 9:23), in which position e is not found. Words with /f/ are spelled with h in the manuscript: hecho (1:5); hambre (6:31, 7:22, 11:14-15, 11:19); hermosa (1:3, 3:13; hermosura, 1:21), hallar (1:8, 1:18; hallo, 1:9; hallado, 1:4, 1:28). /H/ is usually not written: this includes umano (1:14) and all forms of aver. However, we find it on such Latinisms as habituando (2:24) and honestidad (1:21; also onesta, 9:29). There is one example of a word spelled with a superfluous, obviously silent h: horden (14:1; hordenando, 14:2; hordeno, 14:5). All of this suggests an author with vacilating use of h, but usually writing /f/ differently from /h/.
      The phonetic evidence, then, supports the authenticity of the fragment.

FLORIDA STATE UNIVERSITY

     ⁴² In historical linguistics, the tendency of religious language to retard similar secular phenomena is well documented.

Fred Jehle jehle@ipfw.edu	Publications of the CSA	H–Cervantes
URL: http://www.h-net.org/~cervantes/csa/articf90/consonan.htm