Bulletin of Hispanic Studies

The Spread of the Velar Insert /g/ in Medieval Spanish Verbs

For Peer Review

Journal: Bulletin of Hispanic Studies

Manuscript ID: 05-09-BHS-0063.R1

Manuscript Type: Original Article

velar insert, lexical diffusion, Medieval Spanish verbs, analogical Keywords: change

Several verbs in Modern Spanish display a stem-final /g/ in first- person singular of the present indicative and all forms of the present subjunctive. The consonant is not part of the infinitive of these verbs and hence is said to be “inserted” in the conjugation. These verbs include asir, caer, oír, poner, salir, tener, traer, valer, and venir, as well as raer and roer which have /g/ optionally. While the velar element is attested from the earliest written records in tengo, vengo, and pongo, the adoption of /g/ in the remaining verbs occurred over an extended period of time. The phenomenon likewise affected other verbs, some currently obsolete and some that no longer contain /g/. This study contributes to our understanding of the velar insert by documenting its appearance in a corpus of 13th-, 14th-, and 15th-century Spanish texts and viewing its spread through the lens of the theory of lexical diffusion. Abstract: Varios verbos del español moderno contienen una /g/ al final de la raíz en la forma 1 del presente de indicativo y en todas las formas del presente de subjuntivo. La consonante no forma parte del infinitivo y por lo tanto se dice que es “insertada” en la conjugación. Estos verbos incluyen asir, caer, oír, poner, salir, tener, traer, valer y venir, además de raer and roer con /g/ opcional. Mientras que el elemento velar se documenta desde los primeros momentos en tengo, vengo y pongo, la adopción de /g/ en los verbos restantes ocurrió a lo largo de mucho tiempo. El fenómeno también afectó otros verbos, unos que han caído en desuso y unos que ya no contienen /g/. Este estudio contribuye a nuestro conocimiento de la inserción velar con su documentación en un corpus de textos españoles de los siglos XIII, XIV y XV y viendo su expansión a través de la perspectiva de la teoría de la difusión léxica.

The University of Liverpool, Liverpool L69 7ZN Page 1 of 51 Bulletin of Hispanic Studies

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 For Peer Review 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 2 of 51

1 2 3 The Spread of the Velar Insert /g/ in Medieval Spanish Verbs 4 5 6 1. Introduction 7 8 Many verbs in Modern Spanish display a stem-final /g/ in first-person singular of the 9 10 11 present indicative and all forms of the present subjunctive. Since the consonant is not present in 12 13 the infinitive of these verbs, it is said to be inserted in the conjugation and hence has been termed 14 15 the velar insert. These verbs include asir , caer , oír , poner , salir , tener , traer , valer , and venir , 16 17 18 whose first-person indicativeFor and subjunctivePeer forms Review are thus asgo , asga ; caigo , caiga ; oigo , oiga ; 19 20 pongo , ponga ; salgo , salga ; tengo , tenga ; traigo , traiga ; valgo , valga ; and vengo , venga . The 21 22 insertion of /g/ is optional in the verbs raer and roer whose first-person indicative forms can 23 24 25 therefore vary: rayo~raigo ; roo~royo~roigo . In addition to these, other Modern Spanish verbs 26 27 also exhibit a velar element that is absent from the infinitive but nonetheless present in the same 28 29 paradigmatic forms, e.g. decir and hacer : digo , diga ; hago , haga . To these two verbs one may 30 31 32 add yacer which has no less than three conjugational patterns in the present paradigms: 33 34 yazco~yazgo~yago, yazca~yazga~yaga (cf. Elson 1988). While the velar /g/ in digo , hago , and 35 36 Ī Ō Ō Ō 37 yago , which develop from Latin D C , FAC(I) , and IAC(E) , is etymological, in the other 38 1 39 group of verbs it is not. Much discussion has gone into attempting to explain the origins of the 40 41 non-etymological /g/ in verbs such as tengo , vengo , and pongo , however, the spread of the velar 42 43 44 insert from these verbs to the other non-etymological forms mentioned above has not been 45 46 thoroughly documented. This study will contribute to our understanding of the velar insert by 47 48 49 1 50 Malkiel (1974: 309-310), on the other hand, believes that * faço and * yaço would have been the 51 52 expected evolutions and that fago and yago are analogical. He proposes that the forms were 53 54 remodelled through influence of digo~dizes because of speakers’ desire for a more stark contrast, 55 56 57 the voiceless vs. voiced contrast of the affricate pair ç~z (e.g. * yaço~yazes ) being insufficient. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 3 of 51 Bulletin of Hispanic Studies

2 1 2 3 documenting its appearance in a corpus of 13th-, 14th-, and 15th-century Spanish texts and 4 5 6 viewing its spread through the lens of the theory of lexical diffusion. 7 8 1.1 Origins of analogical /g/ 9 10 11 Before describing the methodology of the present study, it would be beneficial to look at 12 13 the development of verbs with stem-final /g/ in Old Spanish to provide some background to the 14 15 problem. Given the variable fate of the development of Latin velar consonants before front vs. 16 17 18 back vowels, many -er For and -ir verbs Peer in Old Spanish Review displayed a velar /g/ in first-person singular 19 20 of the present indicative and the entire present subjunctive and a non-velar element in the rest of 21 22 the present paradigm. Thus we see that in decir , hacer , and yacer , which were mentioned above, 23 24 25 and in others that displayed vowel + /k/ in Latin, e.g. COQ(U) Ō, ADD ŪCŌ, the /k/ voiced 26 27 before the /o/ of the first-person-singular indicative ending and the /a/ of the present subjunctive 28 29 endings, thus producing digo , diga vs. dizes ; fago , faga vs. fazes ; yago , yaga vs. yazes ; cuego , 30 31 32 cuega vs. cuezes ; adugo , aduga vs. aduzes . Likewise affected were -er and -ir verb forms with 33 34 stem-final /rg/ and /ng/ such as those of esparzir and verbs from Latin -NGERE such as tañer , 35 36 37 ceñir , and plañer~plañir . Regular development of Latin /g/ before /o/ and /a/ vs. /e/ produced the 38 39 same paradigmatic alternation seen in the previous verbs: first-person-singular present indicative 40 41 forms and all six present subjunctive forms displayed /g/ while the other present forms did not, 42 43 44 thus: espargo , esparga vs. esparzes ; tango , tanga vs. tañes . It is worthy to underscore the fact 45 46 that this development only affected -er and -ir verbs in Old Spanish since paradigmatic levelling 47 48 characterized the -ar present conjugations. Witness llegar whose present indicative and 49 50 51 subjunctive forms all display /g/: llego , llegas …; llegue , llegues … (Penny 2002: 177-179). 52 53 In addition to these verbs, the velar insert is attested in the earliest texts in forms of tener , 54 55 venir , and poner whose paradigms thus display the same consonantal alternation with regard to 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 4 of 51

3 1 2 3 the presence or absence of /g/: tengo , tenga vs. tienes ; vengo , venga vs. vienes ; pongo , ponga vs. 4 5 6 pones . Tengo , vengo , and pongo do not evolve etymologically from Latin TENE Ō, VENI Ō, and 7 8 PŌNŌ, but rather exhibit an analogical influence of some kind. While scholars agree that the /g/ 9 10 11 in these forms is due to analogy, there are some disagreements in the explanation regarding the 12 13 impetus and sources of the analogy, as well as on the forms posited prior to the analogical 14 15 change. While most scholars depart from the premise that the base forms for tengo and vengo are 16 17 18 *teño and * veño , thereFor are some thatPeer posit * teno andReview * veno as the primitive forms, including 19 20 Urrutia and Álvarez (1983), Alvar and Pottier (1983), Lenfest (1978, 1993), and Mondéjar 21 22 (1995).2 The point of contention regarding the underlying forms revolves around the variable 23 24 25 results of the development of flexional yod in Spanish verbs. Those who argue for * teno and 26 27 *veno believe flexional yod was lost before palatalization which can be seen, for example, in the 28 29 results of SALIAM > sala and VALE Ō > valo whose Old Spanish forms reveal the loss of the 30 31 32 palatalized element (cf. Lenfest 1993: 635-636). On the other hand, those who support *teño and 33 34 *veño believe flexional yod survived long enough to palatalize the stem-final consonant as is 35 36 37 revealed in Portuguese tenho , venho , in Old Italian tegno , vegno , as well as in Old Provençal 38 39 tenh , venh (Grandgent 1905: 126). Given the parallel results in other Romance languages and the 40 41 42 43

44 2 45 Countless historical grammars have posited * teño , *veño for Spanish tengo , vengo beginning 46 47 with Meyer-Lübke (1923) and continuing through Menéndez Pidal (1968) and Lloyd (1987). 48 49 50 While in his earlier work Penny (1991: 151) expresses some reservations about the existence of 51 52 *veño , he seems more convinced that flexional yod survived long enough to palatalize /n/ in the 53 54 second edition of his historical grammar (2002: 175). Palatalized forms are also posited for 55 56 57 Catalan *tiny , *viny (Badía Margarit 1951: 302). 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 5 of 51 Bulletin of Hispanic Studies

4 1 2 3 need to take them into consideration in the explanation of the velar insert, *teño and * veño 4 5 6 appear to be the superior options. 7 8 The evidence from Portuguese just alluded to shows that the velar insert in tengo , vengo 9 10 11 cannot be traced to a common source in Proto-Romance (Malkiel 1974: 321). It is striking, 12 13 however, that the velar element appears in present-conjugation verb forms in an uninterrupted 14 15 wave of dialects that stretches from Castilian in the west to Italian in the east, being absent from 16 17 18 both the western and easternFor flanks Peer (Portuguese andReview Rumanian, respectively). This can be 19 20 witnessed in present forms of tener and venir in the relevant languages: Castilian/Spanish (tengo , 21 22 tenga ; vengo , venga ), Navarro-Aragonese ( tiengo , tienga ; viengo , vienga ), Catalan ( tinc , tingui ; 23 24 25 vinc , vingui ), Occitan ( teni , tinga ; veni , v enga ), Italian ( tengo , tenga ; vengo , venga ). These forms 26 27 with /g/ contrast with those without /g/ found in Portuguese ( tenho , tenha ; venho , venha ) and 28 29 Rumanian ( Ńin , să Ńin ; vin , să vin ). 3 30 31 32 Returning to Spanish forms tengo , vengo , pongo , there is some variation among scholarly 33 34 attempts to explain the analogical velar, as previously stated. For some, the existence of many 35 36 37 verbs with stems ending in /g/ in first-person singular of the present indicative and all forms of 38 39 the present subjunctive was enough to influence the adoption of /g/ in the same forms of tener , 40 41 venir , and poner . Thus, Lloyd (1987: 164) speaks of mutual analogical influence between forms 42 43

44 3 45 Evidence for Navarro-Aragonese is extrapolated from Alvar (1953: 220), while for Catalan 46 47 Moll (1952) and Badía Margarit (1951) are used. Verb forms for Occitan are taken from Internet 48 49 50 sources including http://www.orbilat.com/Languages/Occitan/Grammar/Verbs/index.html and 51 52 http://pagesperso-orange.fr/babel-site/occitan2.html ; cf. Pérez Saldanya (1995) and 53 54 Klausenburger (1984) who list alternate forms for Provençal. The latter lists alternate forms for 55 56 57 Catalan as well. Rumanian forms are provided by a native informant. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 6 of 51

5 1 2 3 of tener and venir , on the one hand, and forms of verbs that descended from Latin -NGERE, on 4 5 6 the other. In other words, due to the phonetic similarity between taño and *teño , combined with 7 8 the alternation taño ~ tango , speakers would then begin alternating forms * teño ~ tengo . He 9 10 11 believes that it is really only necessary to assume that this alternation and the existence of stem- 12 13 final /g/ in commonly used verbs such digo , fago , and adugo produced ‘mental identification’ of 14 15 the /g/ as characteristic of form 1 of the indicative and all forms of the present subjunctive for 16 17 18 certain verbs. The similarityFor of structure Peer of poner Review and the leader verbs tener and venir resulted in 19 20 its early adoption of the velar consonant (1987: 295-296). Penny (2002: 175, 177-179) follows a 21 22 very similar line of reasoning. 23 24 25 For other scholars, however, the existence of verbs with /g/ in form 1 of the present 26 27 indicative and all forms of the present subjunctive does not provide sufficient explanation for the 28 29 adoption of the velar consonant in tener , venir , and poner . Malkiel (1974: 328) for example, 30 31 32 rejects the influence of verbs like taño~tango , ciño~cingo , plaño~plango , as the source of the 33 34 velar insert. He contends that this is problematic due to the fact that: 1) these were not high- 35 36 37 frequency verbs, and 2) their development ( tango > taño ) goes from etymological to analogical, 38 4 39 precisely the opposite of *teño > tengo . Instead, the author believes that the palatalization of 40 41 stem-final /-n/ of TENE Ō, VENI Ō produced a morphophonemic alternation /ñ/~/n/ in the 42 43 44 paradigm (e.g. * teño ~tienes ) which was weakly contrasted. Given speakers’ preference for 45 46 ‘stark’ contrasts, the /ñ/~/n/ alternation was abandoned in favor of /ng/~/n/ in languages like 47 48 49 4 50 Lenfest (1978, 1993) concurs with this point. He argues that speakers extended the /g/ to tengo , 51 52 vengo not in imitation of tango , etc., but rather to restore ‘phonic substance’ and paradigmatic 53 54 balance; in other words, the CVCCV pattern of tengo would match the phonic weight of the 55 56 57 CVVCV(C) pattern of tiene(s) . For a critique of this theory, see Lloyd (1987: 295). 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 7 of 51 Bulletin of Hispanic Studies

6 1 2 3 Spanish and Italian. This new alternation served to more distinctly mark first-person singular of 4 5 5 6 the present indicative. While Malkiel rejects the idea that analogical imitation was the impetus 7 8 for the change , he recognizes that verbs like tango provided a model for the adoption of /g/ 9 10 11 (1974: 345). More recent studies that focus on the morphological implications of the insert /g/ 12 13 have added to Malkiel’s principal argument of the desire to mark first person. Klausenburger 14 15 (1984), for example, argues that through its adoption, /g/ becomes a morpheme of markedness, 16 17 18 and Pérez Saldanya (1995)For adds thatPeer the verbal augmentReview /g/ has a clear morphosyntactic function 19 20 that allows the formal delimitation of a verbal subclass. In the case of Spanish, the subset 21 22 includes verbs of the second and third conjugations which, through the adoption of the -go 23 24 25 ending in certain verbs, distance themselves from the first conjugation (Elvira 1998: 196; cf. 26 27 Maiden 2001). 28 29 As can be seen by the above, the velar insert, particularly its genesis in tengo and vengo , 30 31 32 has been discussed in some depth. Nevertheless, we are still lacking a thorough analysis of the 33 34 spread of the velar insert in medieval Spanish. Malkiel (1974: 327), writing more than three 35 36 37 decades ago, stated: ‘It might be rewarding to present an exhaustive picture of the rivalry, in 38 39 medieval texts, of salga and sala , suelgo and suelo , tuelga and tuella ’. The fragmentary nature of 40 41 42 5 Malkiel’s idea of the adoption of /g/ as an alternative to palatalization has had a great deal of 43 44 45 resonance in subsequent studies, see Pérez Saldanya (1995), Elvira (1998: 194-195); cf. Maiden 46 47 (2001: 46ff) who rejects phonological motivation for the origin of the velar insert. While 48 49 50 depalatalization is a convincing argument for the genesis of tengo , vengo , and tuelgo , it cannot 51 52 be applied to the case of verbs such as salgo and valgo , at least in Spanish, since the simplified 53 54 results salo and valo are attested in the earliest texts. Klausenburger (1984: 135-136) and 55 56 57 Kuryłowicz (1968: 74) seem to ignore this point. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 8 of 51

7 1 2 3 what was known in this regard at the time of his statement still holds true today. This is reflected 4 5 6 in the frequent general statements of relative chronology such as the fact that the adoption of /g/ 7 8 was later in the verbs caer , traer , and oír than it was in salir and valer (cf. Elvira 1998: 195) or 9 10 11 the overly conservative dating that depicts valgo , caigo , traigo , and oigo as Golden Age 12 6 13 phenomena (cf. Penny 2002: 179). Even studies that do provide a more precise dating of the 14 15 adoption of /g/ in the analogical forms only provide data for a small subset of verbs (cf. Ridruejo 16 17 18 1998). In short, documentedFor evidence Peer of the variation Review between etymological and analogical 19 20 forms in medieval texts has yet to be provided, and therefore, further systematic inquiry into the 21 22 problem is warranted. 23 24 25 2. Methodology 26 27 The purpose of this study is to present data taken from a substantial corpus of medieval 28 29 texts in order to trace the spread of /g/, both in verbs in which the insert survives into the modern 30 31 32 language and in those in which it is eventually lost. To this end, the following methodology was 33 34 employed. First, a number of texts were selected that conformed to certain pre-established 35 36 37 criteria: 1) the texts needed to be available in electronic format in order to facilitate the collection 38 39 of data, and 2) the texts had to be edited using rigorous guidelines in order to ensure that the data 40 41 were trustworthy. The texts chosen are among those published by the Hispanic Seminary of 42 43 7 44 Medieval Studies (HSMS) available in CD-ROM format. Once the body of available texts was 45 46 47 6 The fragmentary nature of the current knowledge on the subject is further underscored by the 48 49 50 fact that some studies seem to ignore the existence of the forms salo , sala alongside valo , vala in 51 52 medieval Spanish; see, for example, Elvira (1998: 195) and Maiden (2001: 49). 53 54 7 The texts are taken from the following sources: Corfis and O’Neill (1997); Herrera and 55 56 57 González de Fauve (1997); Kasten, Nitti, and Jonxis-Henkemans (1997); Nitti and Kasten 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 9 of 51 Bulletin of Hispanic Studies

8 1 2 3 pinpointed, specific texts representative of the 13th, 14th, and 15th centuries were selected, texts 4 5 6 that also provided data pertinent to the linguistic change being analyzed. Faulhaber et al. (1984) 7 8 (BOOST ) was used to find a reliable dating for each text, including both the original production 9 10 11 date (OPDT), or date of the original, and the specific production date (SPDT), or date of the 12 13 copy. For the 13th and 15th centuries, texts were chosen whose OPDT and SPDT fall within the 14 15 same century, unless otherwise noted.8 Given the lower availability of works from the 14th 16 17 18 century, selection criteriaFor were somewhat Peer less rigid, Review and texts with an ODPT and SPDT that fall 19 20 within the 14th century were chosen as well as those that are 14th-century copies of earlier 21 22 works. It is to be noted that each text is dated according to its SPDT in order to ensure a more 23 24 25 conservative estimate of the dating of the phenomena found therein. This contrasts sharply with 26 27 the methodology used in other corpora (see below). A total of 60 texts were selected with a 28 29 roughly equal distribution in both number of texts and collective size for each century. The total 30 31 32 corpus comprises 16 texts from the 13th century, 18 texts from the 14th century, and 26 texts 33 34 from the 15th. The 13th-century body represents 16,213 KB of data, while the 14th- and 15th- 35 36 9 37 century texts total 11,338 KB and 11,944 KB, respectively. 38 39 The methodology chosen was preferred to the use of electronic corpora available on the 40 41 Internet such as the Real Academia Española’s Corpus diacrónico del español ( CORDE ) and 42 43 44 Davies’ Corpus del español ( CdE ). While these resources are undoubtedly useful tools, we found 45 46 47 (1997); and O’Neill (1999). For a description of the editorial guidelines used in the transcription 48 49 50 of HSMS texts, see Mackenzie (1997). 51 8 52 Also included are texts that are translations of earlier works which, accordingly, only have an 53 54 SPDT in BOOST . 55 56 9 57 See Appendix A for the complete list of texts and Appendix B for the list of texts by century. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 10 of 51

9 1 2 3 our method preferable for the present study because 1) the body of texts was transcribed using 4 5 6 the same norms, 2) we were able to better control for the dating of the phenomena encountered. 7 8 The results of a data search of forms of traer with /g/ can serve as an illustration of the latter. 9 10 11 The search for trayg* in CORDE reveals that many forms are apparently documented in the 13th 12 13 century, the majority of which are found in the Fuero de Úbeda and in the Fuero de Cuenca , 14 15 both included in the corpus as 13th-century texts. Since information regarding the manuscript 16 17 18 copy used in the editionsFor from which Peer the data are Reviewtaken is not provided on the website, one must 19 20 consult the editions themselves. Regarding the Fuero de Úbeda , the manuscript copy used in the 21 22 edition is believed to be from the 14th century (see Peset et al. 1979). For the Fuero de Cuenca , 23 24 25 attestations of trayg* are found in a 15th-century copy of the text (see Ureña y Smenjaud 1935). 26 27 Naturally, in neither case should the data be considered as 13th-century evidence. 28 29 In CdE the dating of the search results is more easily arrived at as either the date or 30 31 32 identification of the manuscript copy (library and number) is usually provided. However, the 33 34 data are again grouped by the date of the original copy of the manuscript. For example, forms of 35 36 37 traer with /g/ are found in the following texts originally composed in the 13th century but whose 38 39 SPDT falls within the 14th, 15th, or even early 16th century: Biblia latina (14th-century copy), 40 41 General historia V and Poema de Fernán González (15th-century copies), and Gran conquista 42 43 44 de Ultramar (edition printed in 1503). The methodology employed in the present study provides 45 46 obvious benefits for the accurate dating of the spread of the velar insert in medieval Spanish. 47 48 3. Results 49 50 51 This study has revealed new findings for several verbs, which will be presented in 52 53 following groups: 1) documentation of the spread of the velar insert in toller , salir , and valer ; 2) 54 55 evidence regarding forms of ferir , fallir , and prender with /g/; 3) textual evidence of the spread 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 11 of 51 Bulletin of Hispanic Studies

10 1 2 3 of /g/ to soler , doler , and asir ; 4) documentation of the adoption of /g/ in traer , caer , oír , and 4 5 10 6 huir . The pertinent data resulting from the textual searches for each verb are presented in the 7 8 tables of Appendix C and will be referenced accordingly. 9 10 11 3.1 Spread of velar insert: toller , salir , valer 12 13 Evidence from our corpus of texts suggests that the verb toller < TOLLERE may have 14 15 been one of the first verbs to adopt the velar insert after the leader forms tener , venir , and poner . 16 17 18 The data presented in TableFor 3.1a showsPeer that analogical Review forms with /g/ appear in 14 of the 16 19 20 13th-century texts of the corpus a total of 122 times. In contrast, etymological forms are only 21 22 used twice. In other words, 98% of the occurrences that concern us here (form 1 of the present 23 24 25 indicative and all forms of the present subjunctive) are of the type tolg-~tuelg-. Forms of Old 26 27 Spanish coller < COLLIGERE (later coger ) with /g/, e.g. cuelgo , cuelga , were undoubtedly 28 29 influential in the early adoption of /g/ in toller (Dworkin 1983: 167). Despite the relatively high 30 31 32 incidence of toller in the 13th century, its usage drops off sharply in the 14th- and 15th-century 33 34 texts of the corpus. Relevant forms are found in only eight texts of the 14th century for a total of 35 36 11 37 28 occurrences while the 15th century presents only three forms in one text. 38 39 40 41 42 10 The spread of the velar insert is also attested in the verb retiñir < RETINN ĪRE as can be seen 43 44 45 in the following occurrence: retingan (GE4:1). Given the lone attestation, this verb will not form 46 47 part of the larger analysis. Likewise, although forms like remanga are found in the corpus, they 48 49 50 are not commented upon as remanir , along with tener , venir , and poner , is one of the verbs in 51 52 which the velar insert is attested in the first written texts. 53 54 11 For factors related to the demise of toller , its replacement by quitar , and its limited survival in 55 56 57 the more advanced variant tullir ‘to cripple, maim, paralyze’, see Dworkin (1983). 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 12 of 51

11 1 2 3 Another verb that shows early adoption of the velar insert is salir < SAL ĪRE. Forms with 4 5 6 /g/ appear in 15 of the 16 13th-century texts a total of 137 times while etymological forms such 7 8 as SAL(I) Ō > salo and SAL(I)AM > sala appear in six texts a total of 13 times (see Table 3.1a). 9 10 11 While only one text (EE1) makes exclusive use of etymological forms, 10 texts use forms with 12 13 /g/ exclusively. Variation occurs in five texts, as in these two excerpts from GE4 presented in 14 15 (1):12 16 17 18 (1) For Peer Review 19 20 a. Pongamos` | amiztad en uno. & tre & salamos a un campo & | firmemos la. (GE4 21 22 137v) 23 24 25 b. Torme nta | uos tomara. & dolores como a mugier que | esta departo. Nin salgades a 26 27 labrar. nin an- | dedes carrera. Ca la espada del enemigo & el | so pauor; es en 28 29 derredor dela cibdad. (GE4 79r). 30 31 32 For the 13th century overall, analogical forms predominate with a 91% rate of 33 34 appearance. This percentage increases to 98% in the 14th century as forms such as sala appear in 35 36 37 only two texts a total of two times vs. 82 occurrences of forms with /g/ in 17 texts. The two texts 38 39 that make use of etymological forms are ALX and IJ8, which are 14th-century copies of earlier 40 41 works. Thus, it is not certain if the forms are relics of the earlier rendition of the text or if they 42 43 44 are indeed reflective of 14th-century practices. Nevertheless, Mondéjar (1995: 20) finds sala 45 46 documented as late as 1313, which corroborates the data found here. By the 15th century, the 47 48 49 12 50 In the citations taken from the texts of this corpus, transcription norms have been slightly 51 52 modified; c’ and n~ appear here as ç and ñ. Suppressed letters of abbreviations marked with <> 53 54 in the transcribed text appear here in italicized form. Lines of text appear run-on here, but the 55 56 57 original line separation is marked by the use of a vertical bar |. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 13 of 51 Bulletin of Hispanic Studies

12 1 2 3 adoption of /g/ in first-person singular of the present indicative and all forms of the present 4 5 6 subjunctive appears to be complete as etymological forms are not documented in that period. 7 8 Due to its phonetic similarity with the verb salir , valer < VAL ĒRE also acquires the 9 10 11 velar insert, but it is a later phenomenon in valer as the data will show (see Table 3.1a). Only two 12 13 texts in 13th century (FGN, FNV) show forms with analogical /g/ for a total of 13 occurrences. 14 15 Forms without /g/, etymological forms, are attested for a total of 131 occurrences in 10 texts. 16 17 18 Both 13th-century textsFor with analogical Peer valga also Review display the etymological forms. The variation 19 20 is exemplified in the following two citations from FGN: 21 22 (2) 23 24 25 a. SJ fidalgo p ri`siere Ricom ne. | ho mesnadero. o p re so que | uala .Mil. morauidis de 26 27 otro | Regno. (FGN 7v) 28 29 b. % Mandamos por | fuero q ue esta carta asi rayda ho | emendada en tales logares no n | 30 31 32 ualga . (FGN 26v) 33 34 The incidence of analogical forms does not become significant until the 15th century as 35 36 37 only seven occurrences of forms such as valga are found in the 14th century. While analogical 38 39 forms with /g/ only appear 9% of the time in the 13th-century texts of the corpus, this percentage 40 41 increases to 38% in the 15th century Nevertheless, forms without /g/ still outweigh those with the 42 43 44 insert (23 occurrences for valga , etc. vs. 38 occurrences of v ala , etc.). The data regarding total 45 46 occurrences is underscored by that of usage by text. Of the 19 texts of the 15th century in which 47 48 relevant forms are attested, seven make exclusive use of etymological forms, six exclusively 49 50 51 show analogical forms, while the remaining six show variation of the type vala~valga . 52 53 While it is true that the adoption of /g/ was slower to triumph in valer than it was in salir , 54 55 it is imprecise to describe valgo/valga as Golden Age forms (cf. Penny 2002: 179) as they are 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 14 of 51

13 1 2 3 amply attested by the 15th century. In fact, in Nebrija’s (1492 [1980]: 248) description of verbs 4 5 6 of the second and third conjugation that end in -go ‘[que] no siguen la proporción del infinitivo’, 7 8 we find the following: ‘de valer valgo vales ’. Correas (1625 [1954]: 295) likewise provides 9 10 11 valgo as the first-person-singular present indicative form of valer . Etymological vala is able to 12 13 persist through the 15th century, and even into the 16th, in large part because of its frequent use 14 15 in legal texts and in other formulaic contexts such as in fossilized expressions such as ‘válame 16 17 18 Dios ’. Regarding the latter,For it is highly Peer significant Review that 23 of the 38 occurrences (61%) of 19 20 etymological forms of valer in the 15th-century texts of the corpus are constructions of that type, 21 22 those that appeal to God or other supernatural or superior beings. Nineteen of the 23 examples 23 24 25 are of the type ‘así me vala Dios ’ while the remaining four refer to Allah, Muhammad, or the 26 27 devil. Some examples are provided in (3): 28 29 (3) 30 31 32 a. valala el dia- | blo (C01 34r) 33 34 b. Sy .alaha. me vala. (DAN 19v) 35 36 37 c. que dios te vala (ENC 115r) 38 39 d. Asy me vala maoma (MEL 112v) 40 41 e. assi dios me vala (OLI 35v) 42 43 44 f. si dios me vala (PRS 80v) 45 46 In addition to the analogical influence of tengo , vengo , and pongo in the formation of 47 48 salgo and valgo , many scholars cite the influence of verbs like cuelgo and fuelgo which show the 49 50 51 etymological sequence -lg -. Moreover , Alvar and Pottier (1983: 222) credit the lasting power of 52 53 the /g/ in valgo and salgo as necessary to avoid confusion with balo and salo of the verbs balar 54 55 and salar . Whatever role these factors may have played, the most important influence was 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 15 of 51 Bulletin of Hispanic Studies

14 1 2 3 undoubtedly the /g/ that marked first-person singular of the present indicative and the present 4 5 6 subjunctive paradigm of certain verbs, chief among them analogical tengo , vengo , pongo , as well 7 8 as those that displayed etymological /g/. 9 10 11 As mentioned in the introduction (see 1.1) there has been much debate about the possible 12 13 analogical influence of verbs with etymological /g/ on the development of the velar insert. 14 15 Malkiel (1974) was perhaps the first to call into question the traditional explanation of the 16 17 18 influence of -NGERE Forverbs because Peer of their lower Review relative frequency as well as their eventual 19 20 abandonment of /g/ in the paradigmatic levelling that ensued, e.g. tango > taño . Lenfest (1978, 21 22 1993) shares Malkiel’s point of view. For his part, Mondéjar (1995: 17) rejects not only the 23 24 25 influence of -NGERE verbs due to their limited documentation, but also that of verbs like decir , 26 27 cocer , and aducir because they were ‘escasos’ in the present subjunctive. Ridruejo (1998: 728) 28 29 seconds Mondéjar’s opinion. 30 31 32 In an attempt to shed light on the question of the origin of the velar insert as it relates to 33 34 these verbs, the texts of the corpus were analyzed in order to document the presence of 35 36 37 etymological forms with /g/ in Old Spanish as well as the emergence of analogical forms without 38 39 /g/. The data are presented in Tables 3.1b-3.1e. Here we will concentrate on 13th-century 40 41 documentation. Regarding verbs in -NGERE, several are attested in the 13th-century texts of the 42 43 44 present corpus. They include: ceñir < CINGERE, constreñir < CON + STRINGERE, tañer < 45 46 TANGERE, and teñir < TINGERE. 13 All the 13th-century attestations display etymological /g/; 47 48 49 13 50 Forms of enfeñir < EN + FINGERE, llañer < PLANGERE, and reñir < RINGERE with /g/ are 51 52 also attested in GE4 ( enfingo , llangamos , llangan ) and MOA ( rengan ). Given their low 53 54 frequency in the 13th century, their documentation in the 14th and 15th centuries was not 55 56 57 pursued. These data are not included in the tables of Appendix C. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 16 of 51

15 1 2 3 analogical forms with /ñ/ are not attested, and in fact are not found until the 15th century. Forms 4 5 6 with /g/ are attested in 13 of the 16 13th-century texts for a total of 68 occurrences. 7 8 In addition to verbs that descend from Latin -NGERE, etymological /g/ is also found in 9 10 Ī Ō Ō Ō Ū Ō Ō 11 D C > digo , FAC(I) > fago , IAC(E) > yago , ADD C > adugo , SPARG > espargo , and 12 13 COQ(U) Ō > cuego . In Old Spanish, all these verbs display /g/ in first-person singular of the 14 15 present indicative and the entire present subjunctive. Digo and fago/hago are still conjugated in 16 17 18 this way in Modern Spanish,For while Peer the other forms Review become: (yago) ~yazco ~yazgo ; -duzco (as in 19 20 conduzco ); esparzo ; and cuezo . Nevertheless, in the texts of the present corpus, yago persists, 21 22 while indications of -dugo > -duzco ~-duzgo , espargo > esparzo , and cuego > cuezo~cuezgo do 23 24 25 not become evident until the 15th century (see Table 3.1e). Concentrating on the 13th century 26 27 data, forms of aducir , cocer , esparcir , and yacer with /g/ are found in 14 of the 16 13th-century 28 29 texts for a total of 194 occurrences. If we add this to the 68 attestations of forms of -NGERE 30 31 32 with /g/, the total of etymological forms with /g/ for the 13th century is 262. While it cannot be 33 34 denied that these are by no means high frequency verbs, their paradigms, together with those of 35 36 37 decir and hacer , undoubtedly aided in the identification of /g/ as characteristic of the first-person 38 39 singular of the present indicative and all forms of the present subjunctive. It is thus plausible that 40 41 both -NGERE verbs and verbs like decir , hacer , yacer , and aducir , etc. exerted an analogical 42 43 44 influence on * teño, *veño, and * poño producing a third group, tengo , vengo , and pongo . It is 45 46 most likely the analogical influence of all three groups, then, that extends the velar insert to other 47 48 verbs like salir and valer in the manner of lexical diffusion (see section 4), an additional 49 50 51 important factor in the spread of /g/ to these last verbs being the phonetic similarities shared by 52 53 the nasals (of tengo , vengo , pongo ) and the liquids (of salgo and valgo ). 54 55 3.2 Velar /g/ in ferir , fallir , prender 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 17 of 51 Bulletin of Hispanic Studies

16 1 2 3 In addition to its insertion after the liquid lateral /l/ of salir and valer , the velar consonant 4 5 6 also makes itself felt after the liquid tap /r/ of the verb ferir < FER ĪRE. This phenomenon also 7 8 appears to be early as forms with /g/ are found in some 13th-century texts of the corpus (AST, 9 10 11 FGN, FNV, and MOA) as well as in two texts that are 14th-century copies (CID and IJ8). Forms 12 13 with /g/ total 17 occurrences (see Table 3.2). However, after the 14th century, forms with 14 15 analogical /g/ are not attested in the corpus. While fierga is frequently mentioned by scholars as 16 17 18 one of the verbs affectedFor by the velar Peer insert, usually Review cited in the Poema de Mio Cid , no one has 19 20 yet, to our knowledge, commented on the possible Navarro-Aragonese character of this form. It 21 22 is significant that three of the six texts in which forms with /g/ appear in the present corpus are 23 24 14 25 Navarro-Aragonese: FGN, FNV, and IJ8. In fact, these three texts account for 76% of the 26 27 occurrences of fierga (13 of 17), with only four occurrences remaining, one each in the 28 29 Alfonsine AST and MOA and two in CID. Aragonesisms are known to occur in other works of 30 31 32 Alfonso, so the appearance of firgades in CID would be the only form that wants explanation. 33 34 One of the occurrences of the verb is seen in the following passage: 35 36 37 (4) 38 39 Antes q ue ellos legen a lan no p re sentemos les las lanças | Por vno q ue firgades tres siellas 40 41 yran vazias (CID 21r) 42 43 44 45 46 47 14 For purposes of comparison, the alphabetical indices of several 13th-14th-century Aragonese 48 49 50 sources published in Nitti and Kasten (1997) and O’Neill (1999) were examined. Fierga was 51 52 found in FAR seven times and in PL1 once. (See Appendix A for abbreviations.) FAR has a 53 54 13th-century OPDT and a 14th-century SPDT. PL1 has a 14th-century OPDT and a 15th-century 55 56 57 SPDT. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 18 of 51

17 1 2 3 However, as is characteristic of other examples of early literature in Castilian, the Poema de Mio 4 5 6 Cid shows a mixture of dialect features, among them features that can be considered non- 7 8 Castilian (Pountain 2001: 58). 9 10 11 Another early indication of the spread of the velar insert is the form falgades from fallir < 12 13 FALLERE found once in the Alfonsine text EE2 (see Table 3.2). After the 13th century, the verb 14 15 fallir with /g/ is not attested in the texts of the corpus. However, it appears a number of times in 16 17 15 18 14th-century AragoneseFor texts. FallirPeer with /g/ it wouldReview seem, cannot be considered characteristic 19 20 of Castilian. In EE2 it is probably an Aragonesism. 21 22 The use of velar /g/ in the verb prender < V.L. PRENDERE < C.L. PREHENDERE and 23 24 25 its compounds is another use characteristic of Navarro-Aragonese that is found in early Spanish 26 27 texts. Forms of prender and its compounds displaying /g/ are found 50 times in four texts of the 28 29 present corpus with a 13th-century OPDT: IJ8, FGN, FNV, and NOV (see Table 3.2). All four 30 31 32 texts are written in Navarro-Aragonese. Commenting on the appearance of prencat in the Glosas 33 34 Silenses , Menéndez Pidal (1964: 358) describes the verb as ‘forma navarroaragonesa, extraña al 35 36 37 castellano’. This is corroborated in the fact that forms of prender and its compounds with /g/ are 38 39 not attested in any later texts of the corpus but appear in the following 14th-century Aragonese 40 41 texts with a total of 51 occurrences: CQ1, CQ2, FAR, FYO, GC1, LAT, SEC, TUC, and TUY. 42 43 44 (See Appendix A for abbreviations.) 45 46 The presence of the velar insert in Navarro-Aragonese, as attested in falga , prenga , and 47 48 fierga ties in nicely with other morphological evolutions on the Iberian Peninsula. In the east, not 49 50 51 52 53 54 15 Forms of falga appear a total of 12 times in the following texts: CQ2, FYO, GC1, LAT, PAL, 55 56 57 SEC, and TUC. See Appendix A for abbreviations. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 19 of 51 Bulletin of Hispanic Studies

18 1 2 3 only is the /g/ present in a wider variety of verbs, 16 but it is also present in other tenses due to 4 5 6 normal phonological evolution, as is the case of Catalan vinguissen and Aragonese prenguessen . 7 8 The velar insert is absent from Portuguese as attested in venho , tenho , and ponho . Castilian, it 9 10 11 seems, finds itself in the middle of the two extremes. 12 13 3.3 Spread of the velar insert: soler , doler , asir 14 15 The verb soler < SOL ĒRE also becomes involved in spread of the velar insert, if only 16 17 18 temporarily. AlthoughFor there is not Peer abundant data, Review the few attested occurrences (three) show that 19 20 analogical suelgo , suelga appears to be largely a 14th-century phenomenon (see Table 3.3). 21 22 Suelgo occurs once in ALX and once in IJ8. The form is also found in the 15th-century BAE, 23 24 25 however, its use appears to respond to the needs of rhyming as attested in the following passage: 26 27 (5) 28 29 yo te lo rruego | con tu fuego. maguer juego | no n ay tal. que no n desates | pues no n 30 31 32 fuelgo. com mo suelgo | rreçelando. (BAE 165v) 33 34 In addition to salir and valer , it would seem that the analogical influence of verbs like cuelgo and 35 36 37 fuelgo , and tuelgo in the -er category, would be particularly appropriate for this verb as well as 38 39 for duelgo . The spread of the velar insert to doler < DOL ĒRE is perhaps somewhat later than for 40 41 42 16 Cf. Catalan bec ‘bebo’, crec ‘creo’, dec ‘debo’, encenc ‘enciendo’, entenc ‘entiendo’, escric 43 44 45 ‘escribo’, moc ‘muevo’, molc ‘muelo’, prenc ‘prendo’, responc ‘respondo’, venc ‘vendo’, among 46 47 others (Badía Margarit 1951: 301-302, Moll 1952: 215-216). One of the reasons that the 48 49 50 adoption of the velar is more intense in Catalan is that it often served to distinguish between 51 52 first- and third-person present indicative verb forms which had become homonymic through 53 54 regular phonetic development; these include forms of creure , molre, and prendre (Pérez 55 56 57 Saldanya 1995: 418). 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 20 of 51

19 1 2 3 soler . Forms with /g/ occur twice in IJ8 but are attested six times in two 15th-century texts of the 4 5 6 present corpus, G&G and IMI (see Table 3.3). 7 8 The last verb to be dealt with in this section is the verb asir . There is disagreement 9 10 11 regarding the genesis of asir . Menéndez Pidal (1968: 293) credits Germanic influence and says 12 13 that asir is ‘probablemente del germánico sazjan ’. Corominas and Pascual (1980: s.v.), on the 14 15 other hand, state that asir is derived from asa in that asir originally meant ‘coger por el asa’. 16 17 18 Corominas and Pascual’sFor explanation Peer seems more Review plausible. Searches of the electronic text files 19 20 reveal the incipience of the analogical change of asa > asga . One occurrence of forms with /g/ 21 22 was found: asga in LBT (see 6a). The variation among different manuscript versions of the text, 23 24 25 the Libro de buen amor , is very interesting. Witness these citations: 26 27 (6) 28 29 a. % tomola en su falda E leuola a su casa | puso la cabe el fuego açerca de buena brasa 30 31 32 | rrebeujo la culebra ante q ue la el asga (LBT 17r) 33 34 b. tomola e n la falda & leuola a su casa | pusola çerca del fuego & çerca de buena brasa 35 36 37 | rrebolujo la cueluebra ant e q ue la e(s)[l] asa (LBG 74r) 38 39 c. % Tomola en la falda & leuola a su cas’a | pus’ola çerca del fuego / çerca de buena 40 41 blasa | abiuo la culebra / ante q ue la el as’a (LBS 81r)17 42 43 44 Unfortunately, the dating of the copies is unable to provide more precise information on 45 46 the emergence of asga . The form with analogical /g/ appears in LBT which has an OPDT of 47 48 49 17 50 While the context of the citations appears to involve the verb asar rather than asir , the latter is 51 52 understood to be the infinitive of asa given the fact that subjunctive is required after ante que . In 53 54 all three manuscript copies, present or imperfect subjunctive forms are used after all instances of 55 56 57 the temporal conjunction ante que (Modern Spanish antes (de) que ). 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 21 of 51 Bulletin of Hispanic Studies

20 1 2 3 1330 and an SPDT of 1300-1400. LBG has an OPDT of 1330 and an SPDT of 1389, while LBS 4 5 6 has an OPDT of 1300-1400 and an SPDT of 1400-1450. Suffice it to say that the velar insert 7 8 begins to spread to the verb asir in the 14th century, but variation continues in the 15th and 16th 9 10 11 centuries. This information is at variance with Penny’s (2002: 179) indication that asa > asga is 12 13 a Golden Age phenomenon. Although the variation may have regularized during this time, its 14 15 origin dates to the 14th century. 16 17 18 Of the above mentionedFor verbs Peer (3.1-3.3), theReview velar insert survives into Modern Spanish 19 20 only in salgo ... salga , valgo ... valga , and asgo ... asga . In the verb toller/tullir , while the /g/ was 21 22 adopted early in the present paradigms, the usage of the verb declined sharply in the 14th and 23 24 25 15th centuries. Now it is obsolete in its finite forms. The verb fallir was likewise lost, eventually 26 27 succumbing to the spread of the inchoative infix and is now fallecer (see Kauffeld 2007). 28 29 Besides which, although falga appeared in early Castilian texts, it was probably never really 30 31 32 characteristic of that variety . Indeed, the same may be true for fierga from ferir , as seen in the 33 34 high percentage of forms which can be attributed to Navarro-Aragonese. Both verbs are 35 36 37 characteristic of eastern dialects, being attested frequently in Aragonese texts. The form prenga 38 39 from prender , which is only documented in the texts of this corpus written in Navarro- 40 41 Aragonese, is also characteristic of eastern peninsular varieties. 42 43 44 In the remaining verbs, soler and doler , the /g/ never became fully entrenched in their 45 46 conjugations as it did in salir and valer despite the phonetic similarity shared by the four verbs. 47 48 The relative frequency of the verbs most likely played a role. For doler , the fact that it is not 49 50 51 high-frequency verb probably contributed to its inability to sustain the anomalous /g/. Soler , on 52 53 the other hand, generally occurs more frequently than valer . However, it has a very limited use in 54 55 the present subjunctive, which undoubtedly contributed to its succumbing to paradigmatic 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 22 of 51

21 1 2 3 pressure. 18 The fact that /g/ remains in asir , asgo , asga seems to be the exception, since asir is 4 5 6 not a high-frequency verb either. The change asa > asga may be influenced by the spread of the 7 8 inchoative pattern to verbs like ( cuega > ) cueza > cuezga .19 Although the change may have 9 10 11 begun earlier, in the texts of this corpus we find 15th-century documentation of forms of cuezga . 12 13 Witness these passages from the 15th-century GOR that show variation cuega~cueza~cuezgan in 14 15 a single text: 16 17 18 (7) For Peer Review 19 20 a. % Pero enel comienço pongan tan- | ta agua que abaste para todo el cozimiento & | 21 22 cuega fasta que se aparten los huessos. (GOR 141v) 23 24 25 b. El fisico | dilige nte & solicito tome algu nas destas cosas: o | todas: & cueza las en 26 27 vinagre: (GOR 22v) 28 29 c. sy las fojas se co- | mieren de mañana antes de comer o fueren te m- | pradas en vino 30 31 32 blanco & que el vino sea beui- | do de mañana o que la cuezgan en vino & reci- | ba 33 34 el fumo o que su çumo distilado colado & | tibio se distile enla oreja. (GOR 83r) 35 36 37 Another important factor in the adoption of /g/ in first-person singular of the present indicative 38 39 and all forms of the present subjunctive of asir would have been the desire to avoid homonymic 40 41 clash with forms of asar . Witness, for example, the citations in 6a-6c above. 42 43

44 18 45 Frequency information for these verbs is extrapolated from Juilland and Chang-Rodríguez 46 47 (1964: s.vv.). 48 49 19 50 Malkiel (1974: 332) points out that vacillation of the type conozco~conozgo , plazco~plazgo 51 52 reaches its peak during the time of Cervantes. In these alternations we see the influence of the 53 54 velar insert on the inchoative conjugation. In this corpus of texts, this influence is seen in the 55 56 57 forms luzga(n) (CIS, IMI, PER), produzga (IMI), and rreduzga (CMN) from the 15th century. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 23 of 51 Bulletin of Hispanic Studies

22 1 2 3 3.4 Spread of velar insert: traer , caer , oír , huir 4 5 6 In this part of the paper, we would like to address the spread of the velar insert to verbs 7 8 such as traer , caer , oír , and huir in which the /g/ was added after stem-final yod. But first, let us 9 10 Ī Ī 11 review the evolution of these verbs from Latin AUD RE, FUG RE, CADERE, TRAHERE to 12 13 Old Spanish oír , fuir , caer , traer . 14 15 Flexional yod survived in the verbs oír and huir ; AUDI Ō > OSp. oyo and FUGI Ō > OSp. 16 17 18 fuyo . The verbs CADEREFor and TRAHERE Peer did not Review have flexional yod but were remodelled on 19 20 verbs like oyo and fuyo , CAD Ō > OSp. cayo 20 and TRAH Ō > OSp. trayo . The /ȭ/ of oyo and 21 22 23 huyo spread within their verbal paradigms to the other rhizotonic forms in order to provide a 24 25 26 sharper syllabic division between the stem-final vowel and the vowel of the ending, thus 27 28 avoiding hiatus of the type /óe/, /úe/, which was unprecedented (Penny 1991: 151, 2002: 175). 29 30 Eventually, all these verbs succumb to the influence of the velar insert, and we find the 31 32 33 alternation oigo , oyes , oiga ; huigo , huyes , huiga ; caigo , caes , caiga ; traigo , traes , traiga . The 34 35 spread of the velar insert to these verbs, it would seem, would most likely affect caer or traer 36 37 first and only later oír and huir . This seems a logical supposition given the fact that the stem- 38 39 40 final yod of cayo , caya and trayo , traya reflects the same alternation as stem-final /g/ in say, 41 42 tengo , tenga . In other words, form 1 of the present indicative and all forms of the present 43 44 45 subjunctive of caer and traer contain /ȭ/ while other forms of the present paradigms do not. 46 47 48 Evidence from Nebrija’s Gramática of 1492 regarding conjugations for traer , oír , and 49 50 huir supports this notion. In Book 4, Chapter 3 (‘De la conjugación del verbo’), first-person 51 52 53 54 20 Alternatively, cayo can be seen to have developed from Latin *CADE Ō, from the infinitive 55 56 Ē 57 *CAD RE, thus explaining the presence of the stem-final yod (Lloyd 1987: 297). 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 24 of 51

23 1 2 3 singular of the present indicative conjugations for the -ir verbs oír and huir are listed as oio and 4 5 6 huio (1980: 248). Likewise, for the present subjunctive conjugation of oír we find the following: 7 8 ‘Como oia, oias, oiamos, oiáis, oian ’ (1980: 241). In Chapter 6 (‘De la formación del 9 10 11 indicativo’), Nebrija explains the conjugation of -er and -ir verbs that end in -go ; the first 12 13 example of the list is ‘como de traer, traigo, traes ’ (1980: 248). While Nebrija provides no 14 15 information regarding the conjugation of caer , the evidence uncovered supports the assumption 16 17 18 that in this group, verbsFor with stem-final Peer /a/ (e.g. traerReview) suffer the influence of the velar insert 19 20 before those containing back vowels ( oír and huir ). Let us now look at the data from our corpus 21 22 to see if they also support this hypothesis. 23 24 25 Of the four verbs considered here, traer is the first to appear with the velar insert /g/ (see 26 27 Table 3.4a). The form traygan occurs once in the 13th-century FGN. While the total number of 28 29 forms without /g/ are not tallied for the 13th century, in same text forms like traya occur seven 30 31 32 times. The following passages from FGN show the variants trayan and trayga : 33 34 (8) 35 36 37 a. pues ta n bie n t er ne pyno et aq ue illos q ue no an traydo poder q ue lo | trayan luego. 38 39 (FGN 116v) 40 41 b. % Et si el arbol fu- | ere robre ho otro arbol q ue ningu na | fruyta no n trayga puede el 42 43 44 seyn- | erpo del arbol puede defender en- | tegrame nt. (FGN 100v) 45 46 In the 14th century, forms with /g/ occur more frequently, in eight texts for a total of 32 47 48 occurrences. However, forms without /g/ still dominate as they occur in 14 texts a total of 73 49 50 51 times. Of the 16 texts in which relevant forms are found, two show exclusive use of forms with 52 53 /g/ (12.5%), eight show exclusive use of forms without /g/ (50%), while the remaining six show 54 55 variation (37.5%). In the 15th century, forms with /g/ occur much more frequently as they appear 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 25 of 51 Bulletin of Hispanic Studies

24 1 2 3 in 20 texts for 96 occurrences. Forms without /g/ have a higher total incidence, occurring in 13 4 5 21 6 texts 111 times. However, if we extract the data from GOR from the tally, forms with /g/ 7 8 dominate 95:60. Regarding usage by text, of the 24 texts in which relevant forms are found, 11 9 10 11 texts use forms with /g/ exclusively (45.8%), four texts use forms without /g/ exclusively 12 13 (16.7%), while nine texts exhibit variation (37.5%). Note that although variation has not been 14 15 eliminated, the exclusive use of forms with /g/ is significantly higher in the 15th century than in 16 17 18 the 14th century. Likewise,For the exclusive Peer use of oldReview forms has significantly declined from one 19 20 century to the next. 21 22 In explaining the origin of traigo , Malkiel (1974: 336) states, ‘the coexistence of trago 23 24 25 and trayo in neighboring dialects was apt to lead to some such compromise form as traigo ’. 26 27 Lloyd (1987: 297), taking his cue from Malkiel, clarifies this idea by saying that trayo and trago 28 29 were competing standard forms in medieval Castilian. Based on these statements, the form trago 30 31 32 and its variants were searched in the present corpus. A total of 11 forms were found in the 13th 33 34 and 14th centuries: four in FJZ, four in ALX, and three in LBT. These forms would apparently 35 36 37 evolve from a Vulgar Latin form *tragere (Grandgent 1962: 175). Hanssen (1913: 102) and 38 39 Staaff (1907: 314) consider the verbs trago , traga to be Leonese forms. Indeed, Modern 40 41 Portuguese trago from trazer (OPtg. trager )22 together with the opinions of Hanssen and Staaff 42 43 44 would seem to indicate that these verb forms are characteristically Western Iberian. 45 46 47 48 49 21 50 Forms without /g/ occur in GOR a total of 51 times, which perhaps skews the data. 51 22 52 Williams posits V.L. * tracere > trazer and * traco > trago for the Modern Portuguese forms. 53 54 In Old Portuguese, trager existed alongside trazer . While the former has disappeared, it was the 55 56 57 preferred form, trazer being perhaps the more popular variant (1968: 238). 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 26 of 51

25 1 2 3 Returning to our attested forms, the highest number of occurrences in a single text is four, 4 5 6 found in both FJZ and ALX. The appearance of Leonese forms in the first text, Fuero Juzgo , 7 8 would hardly be surprising given the fact that it was translated into the vernacular in the reign of 9 10 11 Fernando III (Jonxis-Henkeman and Craddock 1992: i). The second text, Libro de Alexandre , is 12 13 known to be rife with Leonesisms, particularly in Manuscript O, the same manuscript from 14 15 which these data were obtained. Trago also appears in LBT three times. But rather than the 16 17 18 alternation trago~trayoFor that Malkiel Peer and Lloyd referReview to, in this text trago alternates with traygo 19 20 as we see in this excerpt presented in (9): 21 22 (9) 23 24 25 % yo en mj espynaz’o les trago mucha leña | traygo les la faryna q ue come n de la açeña 26 27 (LBT 21r) 28 29 So, rather than the competing forms mentioned earlier, in Castilian all evidence points to the 30 31 32 merely sporadic use of trago . As such, it cannot be considered influential in the analogical 33 34 change trayo > traigo . 35 36 37 The next verb to be analyzed is caer . Forms of caer with the velar insert are not 38 39 documented in the texts of this corpus until the 14th century (see Table 3.4a). This seems to 40 41 indicate that the spread of /g/ to caigo is slightly later than for traigo . Nevertheless, Menéndez 42 43 44 Pidal (1964: 358) finds kaigamus in the Glosas Emilianenses , which are believed to be from the 45 46 10th or 11th century. Although this form has been disputed, with many scholars believing that 47 48 49 the orthographic g represents a palatal /ȭ/ as it apparently does in the attestation of segamus in 50 51 52 another gloss on the same text (cf. Ridruejo 1998: 729), this form may indeed be a valid 53 54 attestation of the early insertion of the velar. Given the purported Riojan origin of the Glosas , 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 27 of 51 Bulletin of Hispanic Studies

26 1 2 3 this information would also coincide with a dialectal trait that we have seen in several other 4 5 23 6 eastern texts, the prevalent use of the velar insert. 7 8 Returning to the texts of this corpus, forms with /g/ are attested a total of seven times in 9 10 11 IJ8. While the total number of forms without /g/ are not tallied for the entire 14th century, in this 12 13 text itself they are attested and actually predominate as caya and its variants appear 11 times. 14 15 These two passages from IJ8 illustrate the variation: 16 17 18 (10) For Peer Review 19 20 a. por nuestro grado perezcriemos et no | nos leuasse dios aesta t ier ra q ue no cayam - | 21 22 os en cuchieillo/ et nuestras mug er es | & n uest ros fijos sean leuados catiuos/. (IJ8 23 24 25 20r) 26 27 b. Guarda q ue | no caygas en maiamie nto de gafedat | mas q ue fagas q ua`nto mostraren 28 29 los sac er - | dot es del linage de leui/ segu nt lo q ue ma nde | aeillos et cumple | lo. (IJ8 30 31 32 46r) 33 34 In the 15th century, forms with /g/ are much more prevalent. They appear in 11 texts for a 35 36 37 total of 48 occurrences. Forms with /g/ predominate numerically as forms without /g/ are found a 38 39 total of 32 times in 13 texts. Of the 17 texts in which relevant forms are attested, four use forms 40 41 with /g/ exclusively (24%), six use forms without /g/ exclusively (35%), variation between caya 42 43 44 and cayga is found in seven of them (41%). Contrasting these numbers for those of traer , we see 45 46 47 23 Although this topic goes beyond the scope of the present study, it would be worthwhile to 48 49 50 investigate possible eastern influence in the spread of the velar insert in Castilian through the 51 52 analysis of pre-Alfonsine and pre-Heredian texts. In his study of the -ie imperfect using these 53 54 types of documents, Imhoff (2000) finds evidence that suggests a more bilateral dialect contact 55 56 57 between the two geographic variants than what has been traditionally believed. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 28 of 51

27 1 2 3 that forms of caer with /g/ occur more often as a percentage of total relevant forms attested as 4 5 24 6 60% of forms attested for caer have /g/ vs. 46% for forms of traer with /g/. However, exclusive 7 8 use of forms like traigo by text in the 15th century is higher than exclusive use of caigo , 46% vs. 9 10 11 24%, respectively. Additionally, overall frequency of forms with /g/ is higher for traer than for 12 13 caer , 96 vs. 48. In other words, forms like traigo occur twice as often as forms like caigo . 14 15 Likewise, since the usage of forms with /g/ appears later in caer than in traer in this corpus, 16 17 18 overall the evidence suggestsFor that Peertraer is the leader Review form. 19 20 Forms of the verb oír with the velar insert also appear in the 14th century, in two texts for 21 22 a total of 12 occurrences (see Table 3.4a). These two texts (AC5, IJ8) show variation with forms 23 24 25 without /g/. In the 15th century, analogical forms with the velar insert /g/ appear much more 26 27 frequently, being attested in 11 texts for a total of 36 occurrences. Etymological forms dominate 28 29 overall, however, as they are attested a total of 51 times in 14 texts. Of the 21 texts in which 30 31 32 relevant forms are found, seven use analogical forms exclusively (33%), 10 use etymological 33 34 forms exclusively (48%), while the remaining four show variation (19%). The variation between 35 36 37 forms with /g/ and those without /g/ can be appreciated in these two passages from CBO: 38 39 (11) 40 41 a. Agora lo q ue | yo fablo a la t ie rra rruego q ue oya n | las palabras de mj boca. (CBO 42 43 44 79r) 45 46 b. E q ui en oreJas tyene oyga | E por obra bye n lo ponga q ue yo | mucho mas me 47 48 Alargara (CBO 30r-30v) 49 50 51 52 53 54 24 If data from GOR are extracted from the results for both verbs, the percentage of forms with 55 56 57 /g/ is 69% for caer and 61% for traer . 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 29 of 51 Bulletin of Hispanic Studies

28 1 2 3 The exclusive use of forms with /g/ in the verb oír in texts from the 15th century (33%) is higher 4 5 6 than the 24% revealed for caer in the same century. However, attachment to old forms is 7 8 stronger in oír than in caer as 48% of the 15th-century texts make exclusive use of forms like 9 10 11 oya vs. 35% for forms like caya . Likewise, analogical forms with /g/ occur more frequently and 12 13 with a higher percentage of total attested forms in caer vs. oír . Of the 80 total relevant forms 14 15 attested for caer , 48 have /g/ (60%). Relevant forms of oír occur 87 times; 36 of those 16 17 18 occurrences display /g/For (41%). Peer Review 19 20 The last verb to be analyzed in this study is the verb huir , which appears seven times with 21 22 analogical /g/ in the 14th century in the text IJ8 (see Table 3.4a). In this same text, five 23 24 25 occurrences of etymological fuya are also attested. In the 15th century, huir appears with the 26 27 velar insert in seven texts for a total of 10 occurrences. Of these seven texts, five show variation 28 29 with etymological huyo , huya , which dominate overall. Note the variation in C01 presented in 30 31 32 (12): 33 34 (12) 35 36 37 a. Por dios que huygas de ser traydo | en lenguas: que al muy deuoto llaman ypocrita: 38 39 (C01 64v) 40 41 b. Par meno bien hablas. en mi coraçon estas. assi se haga. hu - | yamos la muerte que 42 43 44 somos moços. (C01 74r) 45 46 The high incidence of variation with huya in 15th-century texts which display huiga , in addition 47 48 to the lower overall frequency of analogical forms, points to the unstable position the velar insert 49 50 51 held in this verb. In fact, in Modern Spanish, forms with /g/ do not survive in the standard. Why 52 53 is this so? 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 30 of 51

29 1 2 3 Well, we can be certain that the reason is not as Malkiel (1974: 339) suggests because 4 5 6 ‘huigo appeared on the scene far too late to be rammed through a network of defenses available 7 8 to conservative speakers and writers of the 17th century’. As shown in the data, huigo appears at 9 10 11 roughly the same time as oigo . Moreover, as evidenced in Correas’ grammar, huigo was 12 13 considered an acceptable variant of huyo as late as the first quarter of the 17th century (1625 14 15 [1954]: 299). A more appropriate explanation for the demise of huigo is that perhaps, as Lloyd 16 17 18 (1987: 353) points out,For oigo ‘I hear’ Peer would have supportReview from digo ‘I say’. Conversely, huyo (as 19 20 opposed to huigo ), would be supported by semi-learned verbs in -tribuyo , -stituyo and arguyo 21 22 (Malkiel 1974: 338). 23 24 25 In addition to the information provided in the previously analyzed verbs, the analysis of 26 27 this corpus of texts also provides glimpses of other verb forms in which /g/ is inserted after stem- 28 29 final yod (see Table 3.4b). The influence of the /g/ on verbs like creer , leer , and reír and its 30 31 32 compounds is seen in the attested forms leyga (IJ8) from the 14th century and sonrrygo (BAE) 33 34 from the 15th. The influence of the velar insert on -uir verbs other than huir is seen in the form 35 36 25 37 destruyga (IJ8) as early as the 14th century. The beginning of the extension of /g/ to raer , 38 39 40 41 42 25 The form destruyga is also found in the 15th-century text TR2, which does not form part of the 43 44 45 corpus. Evidence from this text sheds light on the question of the lack of triumph of the velar 46 47 insert in learned verbs ending in -uyo . In TR2 numerous forms in hue- (standard fue-) are also 48 49 50 found (e.g. huera , huesemos ). In contrast, in TR1, the copy of the text selected to form part of the 51 52 general corpus, the form of destruir appears as destruya (without /g/) and all perfect forms of 53 54 ser/ir appear with fue- except in one instance. Since /f/ had become the standard urban 55 56 57 pronunciation before /ue/ by the 15th century (Penny 2002: 92-93), this suggests that forms such 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 31 of 51 Bulletin of Hispanic Studies

30 1 2 3 which is now accepted in the standard, is also witnessed as early as the 14th century in rayga 4 5 6 (IJ8) as well as in the 15th century in rraygas (CMN). No examples were found of the influence 7 8 of the velar insert on royo , haya , or vaya , which may be later phenomena. 9 10 11 4. Model of lexical diffusion 12 13 The theory of lexical diffusion proposes that while sound change is abrupt, the diffusion 14 15 of the change in the speaker’s vocabulary is gradual. In other words, a change arises in a small 16 17 18 subset of morphemes andFor affects thePeer relevant morphemes Review containing the sound in succession 19 20 (McMahon 1994: 50, Wang 1969).26 Although the diffusionist theory was originally formulated 21 22 to explain sound change, it can be applied to morphological change since the gradual view of 23 24 25 linguistic change is not applicable to those of a non-phonetic nature, which can only proceed 26 27 from X to Y with no intervening steps. The similarities between this model of sound change and 28 29 other non-phonetic changes have been observed for some time; cf. Wang (1969: 17-18). Writing 30 31 32 at the early part of the last century, Sturtevant, for example, observes: ‘[The] spread of a sound 33 34 change from word to word closely resembles analogical change; the chief difference is that in 35 36 37 analogical change the association groups are based upon meaning, while in this case the groups 38 39 are based upon form’ (1961: 80). 40 41 Returning to the topic that concerns us here, we have seen that the adoption of the velar 42 43 44 insert proceeded along the lines of lexical diffusion. The leader forms tengo , vengo , and pongo 45 46 adopted the /g/ in preliterary times. The innovation then spread to other verbs that also belonged 47 48 49 50 as destruiga were characteristic of more popular speech, and not of the standard language, thus 51 52 supporting the maintenance of the -uyo ending in these learned verbs. 53 54 26 This contrasts with the Neogrammarian model that proposes that regular sound change is 55 56 57 phonetically gradual and lexically abrupt. For more, see Wang (1969), McMahon (1994: 50-68). 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 32 of 51

31 1 2 3 to the second and third conjugations and shared other traits with the leader verbs. While tengo , 4 5 6 vengo , and pongo had a nasal in syllable-final position, the next major verb affected had a liquid 7 8 /l/ in that position (e.g. salgo ), nasals and liquids occupying adjacent positions on the scale of 9 10 11 sonority in Spanish (Hualde 2005: 71-72) . The change then spread to other verbs with a liquid in 12 13 syllable-final position (e.g. fiergo ), and still later to others that also contained a continuant, e.g. 14 15 the glide /j/ as in traigo or /s/ as in asgo . In fact, if one consults a list of the 500 most frequent 16 17 18 words in the Spanish languageFor and Peer the -er and -ir Review verbs that occur among them, the velar insert 19 20 can be seen to have affected a large portion of verbs that exhibit this phonetic structure, i.e. a 21 22 stem-final continuant in form 1 of the present indicative and all forms of the present 23 24 27 25 subjunctive. While in some the velar has persisted (e.g. tengo , vengo , pongo , salgo , valgo , 26 27 traigo , caigo , oigo ), in others it was eventually lost or never became fully entrenched in the 28 29 paradigm in Castilian (e.g. soler , suelgo ; creer , creigo ; leer leigo ). The only popular verbs of the 30 31 32 list that exhibit the same structure but which do not adopt the velar insert are querer , morir , and 33 34 correr . In all three instances, the forms with /g/ would have displayed an alveolar tap in stem- 35 36 37 final position. Given the eventual demise of fiergo in Castilian, this phonetic context may not 38 39 have been favorable to the adoption of /g/ in that variety. An additional factor in the case of 40 41 42 27 See Juilland and Chang-Rodríguez (1964: 385-396). Many of the most frequently used -er and 43 44 45 -ir verbs display some form of irregularity in first person, e.g. verbs with final /j/ ( soy , voy ), 46 47 verbs that follow the inchoative pattern ( parezco , conozco , produzco , ofrezco , nazco ), verbs with 48 49 50 an etymological /g/ that is not present in the infinitive, e.g. hago , digo , and verbs with the velar 51 52 insert. It is interesting to note that the only verbs in Modern Spanish containing the velar insert 53 54 that do not figure among the 500 most frequent words of the Spanish language are obsolescent 55 56 57 asir and variable raer and roer . 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 33 of 51 Bulletin of Hispanic Studies

32 1 2 3 correr and morir could have been their relatively infrequent use in the pertinent verb forms, i.e. 4 5 6 first-person present indicative and present subjunctive (see Juilland and Chang-Rodríguez 1964: 7 8 s.vv.). Alternatively, the change may simply have ceased to be operative before these verbs were 9 10 11 affected. 12 13 To view the chronological progress of the lexical diffusion of the insertion of /g/ in the 14 15 verbs treated in our analysis (3.1-3.4), we can see that in the 13th century, the following verb 16 17 18 forms with /g/ were documentedFor (conflatedPeer here forReview convenience in first-person-singular 19 20 indicative form): tuelgo , salgo , falgo , and fiergo in Castilian texts (four verbs), as well as valgo , 21 22 prengo , and traigo in those of Navarro-Aragonese. In the 14th century, the /g/ is attested in a 23 24 25 greater number of verbs: tuelgo , salgo , valgo , fiergo , asgo , suelgo , traigo , and oigo in Castilian 26 27 texts (eight verbs), as well as prengo , duelgo , caigo , and huigo in Navarro-Aragonese. In the 28 29 15th century, the velar insert is attested in a still greater number of verbs in Castilian (nine): 30 31 32 tuelgo , salgo , valgo , suelgo , duelgo , traigo , caigo , oigo , huigo . This pattern follows what Wang 33 34 (1969: 16) proposes to be true of lexical diffusion: ‘As the change continues to operate, an 35 36 37 increasing portion of the relevant morphemes will become affected’. However, it is also true that 38 39 the longer a change takes to run its course, the larger the likelihood that a competing change will 40 41 be encountered (1969: 11). 42 43 44 Competing forces could plausibly include a decrease in the overall number of -er and -ir 45 46 verbs with /g/. As discussed above (3.1), while verbs such as those that develop from Latin - 47 48 NGERE as well as aducir , cocer , and esparcir persist with etymological /g/ until the 15th 49 50 51 century, their frequency decreases from century to century (see Tables 3.1b-3.1d). Likewise, 52 53 beginning in the 15th century, analogical forms without /g/ are attested in relatively large 54 55 numbers (see Table 3.1e). As these verbs succumbed to pressure from their own paradigms to 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 34 of 51

33 1 2 3 eliminate the /g/ (or from other conjugational patterns in the case of verbs in -ucir ), they 4 5 6 provided one less model for analogical influence, or conversely, a model for /g/-dropping. 7 8 Additionally, the eventual demise of toller had the effect of further reducing the number of verbs 9 10 11 with -go , -ga . Thus, the velar insert was unable to triumph in certain verbs that it had affected 12 13 (e.g. soler , doler ) and lost steam before affecting other -er and -ir verbs with stem-final 14 15 continuants. 28 The decrease in the pattern stem-final consonant + /g/ in -er and -ir verbs is 16 17 18 probably also the reasonFor that the adoptionPeer of /g/ onlyReview remains productive after the 14th century in 19 20 verbs whose stem does not end in a consonant (e.g. huigo , destruigo , raigo , haiga , vaiga in 21 22 Castilian). The fact that the /g/ remains in digo , hago , and yago , coupled with the example of 23 24 25 traigo , caigo , oigo , provided a continued model of analogy for this subgroup of verbs. 26 27 5. Conclusions 28 29 To summarize, the principal contribution of this study is that of dating the spread of the 30 31 32 velar insert throughout the 13th, 14th, and 15th centuries. The verb salir shows variation 33 34 salo~salgo already in the 13th century, with a preference for salgo . Valer , in imitation of salir , 35 36 37 adopts the /g/ later with a high incidence of variation persisting even through the 15th century. 38 39 The beginning of the spread of /g/ to asir is seen in the 14th century, much earlier than had 40 41 previously been thought. Similarly, the spread of the velar insert to traer , caer , and oír can no 42 43 44 longer be considered a principally post-medieval phenomenon as forms with /g/ are heavily 45 46 47 28 Another competing force against the spread of the velar insert would be the tendency toward 48 49 50 paradigmatic balance in the verbs themselves. While on the one hand, morphological change 51 52 favors forms that are more marked, the desire for uniformity of form and function, on the other 53 54 hand, often suppresses allomorphic variation. For more on these two contrary forces, see Elvira 55 56 57 (1998: 216-227). 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 35 of 51 Bulletin of Hispanic Studies

34 1 2 3 attested in the 14th and 15th centuries with variation persisting longer and to a greater degree in 4 5 6 caer and oír than in traer , the leader form. In fact, it is no longer possible to say that the 7 8 generalization of /g/ is slower in traer than in valer as the adoption of the consonant occurs at 9 10 11 roughly the same time in the two verbs and variation persists in valer longer than in traer . 12 13 Throughout this analysis, we have witnessed a particularly high incidence of the early spread of 14 15 /g/ in eastern texts. While here we have concentrated on the lexical diffusion of the velar insert in 16 17 18 medieval Spanish, the Forpossible geographic Peer diffusion Review of the change from east to west needs to be 19 20 examined in order to more fully explain the relationship between this linguistic trait and the 21 22 development of the Castilian verb forms. 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 36 of 51

35 1 2 3 Appendix A: Electronic Sources 29 4 5 6 In Corfis and O’Neill (1997): 7 8 C01: Celestina, [Burgos: Fadrique de Basilea, 1499?] , New York: Hispanic Society. 9 10 11 In Herrera and González de Fauve (1997): 12 13 CIR: Cirugia rimada , Madrid: Nacional MS. 2153. 14 15 GOR: Bernardo de Gordonio: Lilio de medicina , Madrid: Nacional I-315. 16 17 18 SUM: Sumario de la medicinaFor con Peer un compendio Review sobre las pestíferas bubas , Madrid: Nacional 19 20 I-1169. 21 22 In Kasten, Nitti, and Jonxis-Henkemans (1997): 23 24 25 ACE: Libros de ajedrez, dados y tablas , Escorial: Monasterio T.I.6. 26 27 ALB: Canones de Albateni , Paris: Arsenal 8322. 28 29 AST: Libros del saber de astronomía , Madrid: Universitaria 156. 30 31 32 EE1: Estoria de España I , Escorial: Monasterio Y.I.2. 33 34 EE2: Estoria de España II , Escorial: Monasterio X.I.4. 35 36 37 GE1: General estoria I , Madrid: Nacional MS. 816. 38 39 GE4: General estoria IV , Rome: Vatican Urb. Lat. 539. 40 41 JUZ: Judizios de las estrellas , Madrid: Nacional MS. 3065. 42 43 44 LAP: Lapidario de Alfonso X , Escorial: Monasterio H.I.15. 45 46 LEY: Libro de las leyes , London: British Library Add. 20787. 47 48 MOA: Moamyn-Libro de las animalias , Madrid: Nacional MS. Res. 270. 49 50 51 PIC: Picatrix de Alfonso X , Rome: Vatican Reg. Lat. 1283. 52 53 54 29 Texts whose abbreviation appears in italics did not form part of the principal corpus, but rather 55 56 57 were used for purposes of comparison. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 37 of 51 Bulletin of Hispanic Studies

36 1 2 3 In Nitti and Kasten (1997): 4 5 6 CQ1 : Crónica de los conqueridores I , Madrid: Nacional MS. 2211. 7 8 CQ2 : Crónica de los conqueridores II , Madrid: Nacional MS. 10134 bis. 9 10 11 FAR : Fueros de Aragón , Madrid: Nacional MS. 458. 12 13 FGN: Fuero general de Navarra , Madrid: Nacional MS. 248. 14 15 FNV: Fuero de Navarra (versión C) , Salamanca: Universitaria 1947. 16 17 18 FYO : Flor de las ystoriasFor de Orient Peer, Escorial: Monasterio Review Z.I.2. 19 20 GC1 : Grant crónica de Espanya I , Madrid: Nacional MS. 10133. 21 22 LAT : Livres dou tresor , Gerona: Catedral 20-a-5. 23 24 25 PL1 : Plutarch I , Paris: Nationale Esp. 70. 26 27 SEC : Secreto secretorum , Escorial: Monasterio Z.I.2. 28 29 TUC : Tucidides , Madrid: Nacional MS. 10801. 30 31 32 TUY : Crónica del tudense , Stockholm: Kungliga D 1272a. 33 34 In O’Neill (1999): 35 36 37 AC1: Leyes del estilo , Escorial: Monasterio Z.II.8. 38 39 AC5: Fuero real; Leyes del estilo , Madrid: Nacional MS. 5764. 40 41 ALX: Libro de Alexandre , Madrid: Nacional MS. Vit. 5-10. 42 43 44 BAE: Cancionero de Baena (Dutton PN1), Paris: Nationale Esp. 37. 45 46 BUR: Fuero de Burgos , Philadelphia: Free Library E-MS. 245. 47 48 CBO: Arcipreste de Talavera: Corbacho , Escorial: Monasterio h.III.10. 49 50 51 CD1: Castigos y documentos , Escorial: Monasterio Z.III.4. 52 53 CD2: Castigos e documentos de Sancho IV , Madrid: Nacional MS. 6559. 54 55 CID: Poema de Mío Cid , Madrid: Nacional MS. Vit. 7-17. 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 38 of 51

37 1 2 3 CIS: Arte cisoria , Escorial: Monasterio f.IV.1. 4 5 6 CLV: Claros varones de Castilla , Madrid: Nacional I-1569. 7 8 CMN: Arte complida de cirugía (Cirugía menor) , Madrid: Nacional MS. 2165. 9 10 11 CUZ: Libro del kuzari , Madrid: Nacional MS. 17812. 12 13 DAN: Danza de la muerte , Escorial: Monasterio b.IV.21. 14 15 DIV: Tratado de adivinar y de magia , Madrid: Nacional MS. 6401. 16 17 18 DON: Libro de las donasFor, Escorial: Peer Monasterio h.III.20.Review 19 20 ENC: Cancionero de las obras de Juan del Encina (Dutton 96JE), Madrid: Academia Española 21 22 I-8. 23 24 25 FJZ: Fuero Juzgo , New York: Hispanic Society B2567. 26 27 G&G: Grimalte y Gradissa , Madrid: Nacional I-382. 28 29 HER: Los doze trabajos de Hércules , Madrid: Nacional MS. 27. 30 31 32 IJ8: Biblia I.I.8 , Escorial: Monasterio I.I.8. 33 34 IMI: Imitatio Christi , Madrid: Nacional I-977. 35 36 37 LBG: Libro de buen amor -G, Madrid: Academia Española 19. 38 39 LBS: Libro de buen amor -S, Salamanca: Universitaria 2497. 40 41 LBT: Libro de buen amor -T, Madrid: Nacional MS. Vit. 6-1. 42 43 44 LCB: Libro de los caballos , Escorial: Monasterio b.IV.31. 45 46 LEO: Leomarte: Sumas de la historia troyana , Madrid: Nacional MS. 9256. 47 48 MEL: Historia de la linda Melosina , London: British Library IB.42463. 49 50 51 MON: Libro de la montería , Escorial: Monasterio Y.II.19. 52 53 NOV: Fuero de la Novenera , Salamanca: Universitaria 2652. 54 55 OLI: Enrique fi de Oliva , Vienna: Oesterreichishe Nationalbibliothek Ink. G.5.39. 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 39 of 51 Bulletin of Hispanic Studies

38 1 2 3 PAL : Libro de Palladio , Madrid: Nacional MS. 10211. 4 5 6 PER: Mostrador e enseñador de los turbados (perplejos ), Madrid: Nacional MS. 10239. 7 8 PN5: Cancionero castellano de París (Dutton PN5), Paris: Nationale Esp. 227. 9 10 11 PRS: Cancionero de París (Dutton PN12), Paris: Nationale Esp. 313. 12 13 REY: Crónica de once reyes , Escorial: Monasterio Y.I.12. 14 15 SL3: Guerra de Jugurtha , Escorial: Monasterio g.III.11. 16 17 18 SPC: Espéculo , Madrid:For Nacional PeerMS. 10123. Review 19 20 TR1: Triunfo de amor , Madrid: Nacional MS. 22019. 21 22 TR2 : Triunfo de amor , Seville: Colombina 5-3-20. 23 24 25 TRS: Cuento de Tristán de Leonis , Rome: Vatican 6428. 26 27 ULT: Gran conquista de Ultramar , Madrid: Nacional MS. 1187. 28 29 VAS: La estoria del noble Vaspasiano , London: British Library IA.53510. 30 31 32 Y89: Ysopete ystoriado , Escorial: Monasterio 32-I-13. 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 40 of 51

39 1 2 3 Appendix B: Electronic Sources by Century 4 5 30 6 13th Century: ACE, ALB, AST, BUR, EE1, EE2, FGN, FJZ, FNV, GE1, GE4, JUZ, LAP, 7 8 LEY, MOA, PIC 9 10 31 11 14th Century: AC1, AC5, ALX, CD1, CD2, CID, IJ8, LBG, LBS, LBT, LCB, LEO, MON, 12 13 NOV, REY, SPC, TRS, ULT 14 15 14th-century Aragonese texts: CQ1, CQ2, FAR, FYO, GC1, LAT, PAL, SEC, TUC, TUY, PL1 16 17 32 18 15th Century: BAE, C01,For CBO, CIR,Peer CIS, CLV, ReviewCMN, CUZ, DAN, DIV, DON, ENC, G&G, 19 20 GOR, HER, IMI, MEL, OLI, PER, PN5, PRS, SL3, SUM, TR1, TR2 , VAS, Y89 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 30 Although the SPDT of EE2 is not confirmed as definitely falling within the 13th century, 41 42 evidence from this source is included with the 13th-century data given the fact that the text forms 43 44 45 part of the Alfonsine corpus and as such is deemed more trustworthy as a linguistic source 46 47 representative of the 13th century. 48 49 31 50 The SPDT of LBS is 1400-1450, however, it is included with the 14th-century texts due to its 51 52 relationship with LBG and LBT. Relevant variations among the manuscript copies are noted in 53 54 the analysis. 55 56 32 57 The terminus a quo for DON dates from the last decade of the 14th century. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 41 of 51 Bulletin of Hispanic Studies

40 1 2 3 Appendix C: Data 33 4 5 6 Table 3.1a 7 8 Verb Century Texts and occurrences Totals 9 salir 10 th 11 salg- 13 c. ACE:1, ALB:2, AST:20, BUR:11, EE2:14, FGN:14, FJZ:2, 137 12 FNV:11, GE1:17, GE4:12, JUZ:9, LAP:1, LEY:8, MOA:11, 13 PIC:4 14 14 th c. AC1:2, AC5:2, ALX:1, CD2:3, CID:1, IJ8:22, LBG:2, 82 15 LBS:3, LBT:1, LCB:18, LEO:5, MON:4, NOV:2, REY:5, 16 17 SPC:5, TRS:2, ULT:1 th 18 15 c. ForBAE:9, Peer C01:8, CBO:5, Review CIR:11, CIS:4, CLV:1, CMN:11, 99 19 CUZ:7, DAN:1, DIV:1, DON:1, ENC:5, GOR:15, IMI:1, 20 MEL:3, OLI:1, PER:7, PN5:1, PRS:1, SL3:1, SLM:1, TR1:1, 21 VAS:1, Y89:2 22 th 23 sal- 13 c. EE1:1, FJZ:1, GE1:2, GE4:3, LAP:5, PIC:1 13 th 24 14 c. ALX:1, IJ8:1 2 25 15 th c. 0 26 27 28 29 30 31 32 33 34 35 33 The data presented in the tables of this appendix reflect first-person singular of the present 36 37 38 indicative and present subjunctive forms of the verbs analyzed, as well as their compounds. All 39 40 persons and orthographic variants have been conflated under the lemma reflecting the forms with 41 42 /g/ and the forms without /g/. The data were obtained from the electronic files of the respective 43 44 45 text, departing from the alphabetical index. As many verbs forms in Old Spanish shared a 46 47 common spelling, e.g. cayo ‘I fall’ and cayo ‘he/she/it fell’, traya ‘I/he/she/it used to bring’ and 48 49 50 traya ‘that I/he/she/it may bring’, and indeed many verb forms share(d) spellings with words of 51 52 other lexical categories, e.g. sala ‘room’ and sala ‘that I/he/she/it may leave’, the texts 53 54 themselves were analyzed in these cases for context cues indicating the correct interpretation of 55 56 57 the form in order to present the most accurate data possible. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 42 of 51

41 1 2 3 4 5 toller th 6 tolg-~tuelg- 13 c. ALB:2, AST:7, BUR:2, EE2:1, FGN:6, FJZ:4, FNV:4, 122 7 GE1:12, GE4:5, JUZ:5, LAP:3, LEY:7, MOA:62, PIC:2 8 14 th c. CD1:2, CD2:2, IJ8:2, LBG:1, LBS:1, LCB:3, SPC:16, ULT:1 28 9 15 th c. HER:3 3 10 th 11 toll-~tuell- 13 c. GE1:1, MOA:1 2 12 14 th c. 0 13 15 th c. 0 14 valer 15 th 16 valg- 13 c. FGN:8, FNV:5 13 th 17 14 c. CD1:3, IJ8:1, NOV:1, TRS:2 7 18 15 th c. ForBAE:2, Peer CLV:1, CMN:4, Review DON:2, ENC:2, G&G:2, IMI:2, 23 19 MEL:1, PN5:2, PRS:2, SUM:1, TR1:2 20 th 21 val- 13 c. ACE:2, BUR:14, EE1:1, EE2:5, FGN:10, FJZ:73, FNV:9, 131 22 GE1:7, GE4:6, LEY:4 th 23 14 c. AC1:4, AC5:3, ALX:5, CD1:1, CD2:4, CID:20, IJ8:10, 114 24 LBG:4, LBS:3, LBT:1, MON:1, NOV:1, REY:1, SPC:43, 25 TRS:13 26 15 th c. BAE:14, C01:3, CBO:4, CLV:2, CMN:1, DAN:1, ENC:2, 38 27 28 GOR:1, MEL:1, OLI:6, PRS:1, VAS:1, Y89:1 29 30 Table 3.1b 31 32 Etymological forms with Texts and occurrences: 13 th -century Totals 33 34 /g/ 35 -NGERE: 36 ceñir: cing- EE1:2, EE2:1, FGN:2, FNV:2, GE1:1 8 37 -streñir: -string- FGN:1, FJZ:18, LEY:2 21 38 39 tañer: tang- ALB:1, AST:11, BUR:2, EE1:2, EE2:4, GE1:2, GE4:1, 35 40 LAP:3, LEY:2, MOA:7 41 teñir: ting- GE1:2, GE4:1, MOA:1 4 42 Total for –NGERE in 13 th century 68 43 Other: 44 45 aducir: adug- ACE:1, AST:1, BUR:5, EE2:1, FGN:22, FNV:14, 72 46 GE1:10, GE4:14, LEY:4 47 cocer: cog-~cueg- GE1:4, LAP:1, LEY:1, MOA:66 72 48 esparcir: esparg- EE1:1, MOA:1 2 49 50 yacer: yag- BUR:4, EE1:7, EE2:9, FGN:6, FJZ:1, FNV:7, GE1:7, 48 51 GE4:1, JUZ:2, LAP:1, LEY:3 th 52 Total for ‘Other’ in 13 century 194 53 Total occurrences of etymological forms with /g/: 13 th 262 54 century 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 43 of 51 Bulletin of Hispanic Studies

42 1 2 3 Table 3.1c 4 5 th 6 Etymological forms with Texts and occurrences: 14 -century Totals 7 /g/ 8 -NGERE: 9 ceñir: cing- ALX:1, IJ8:1, SPC:1 3 10 11 -streñir: -string- 0 12 tañer: tang- AC1:1, CD1:3, CD2:5, IJ8:4, LBG:3, LBS:3, LBT:1, 59 13 LCB:9, LEO:1, MON:22, SPC:7 14 teñir: ting- 0 15 th 16 Total for –NGERE in 14 century 62 17 Other: 18 aducir: adug- For ALX:1, Peer CID:4, LBG:1, Review LBS:1, LBT:1, NOV:1, SPC:12 21 19 cocer: cog-~cueg- IJ8:4, LCB:3, MON:2 9 20 esparcir: esparg- IJ8:1 1 21 22 yacer: yag- ALX:3, CD2:1, IJ8:11, LBG:2, LBS:3, LBT:1, MON:3, 47 23 NOV:14, REY:3, SPC:5, ULT:1 24 Total for ‘Other’ in 14 th century 78 25 Total occurrences of etymological forms with /g/: 14 th 140 26 27 century 28 29 Table 3.1d 30 31 Etymological forms with Texts and occurrences: 15 th -century Totals 32 /g/ 33 34 -NGERE: 35 ceñir: cing- BAE:1, ENC:2, GOR:2 5 36 -streñir: -string- BAE:2, GOR:2 4 37 tañer: tang- BAE:3, C01:1, CMN:18, GOR:2, PER:1, SL3:1, Y89:1 27 38 39 teñir: ting- GOR:1 1 th 40 Total for –NGERE in 15 century 37 41 Other: 42 -ducir: -dug- CIS:1, PN5:1 2 43 44 cocer: cog-~cueg- CIR:1, CIS:1, ENC:2, GOR:25 29 45 esparcir: esparg- 0 46 yacer: yag- BAE:6, CMN:9 15 47 Total for ‘Other’ in 15 th century 46 48 Total occurrences of etymological forms with /g/: 15 th 83 49 50 century 51 52 53 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 44 of 51

43 1 2 3 Table 3.1e 4 5 th 6 Non-etymological forms Texts and occurrences: 15 -century Totals 7 (formerly with /g/) 8 -NGERE: 9 ceñir: ciñ- ENC:2 2 10 11 -streñir: -striñ- CLV:4, CMN:2, CUZ:1, GOR:4, MEL:1, SUM:1 13 12 tañer: tañ- ENC:2 2 13 teñir: tiñ- GOR:2 2 14 Total for –NGERE in 15 th century 19 15 16 Other: 17 -(d)ucir: -(d)uzc-~-(d)uzg- CIS:1, CMN:1, IMI:2, PER:1 5 18 cocer: coz-~cuez- For ENC:2, Peer GOR:25, SUM:2Review 29 19 esparcir: esparz- CMN:2, CUZ:1, ENC:1 4 20 yacer: yaz- 0 21 th 22 Total for ‘Other’ in 15 century 38 th 23 Total occurrences of non-etymological forms: 15 57 24 century 25 26 27 Table 3.2 28 29 Verb Century Texts and occurrences Totals 30 fallir 31 falg- 13 th c. EE2:1 1 32 th 33 14 c. 0 th 34 15 c. 0 35 ferir 36 fierg-~firg- 13 th c. AST:1, FGN:7, FNV:2, MOA:1 11 37 14 th c. CID:2, IJ8:4 6 38 th 39 15 c. 0 40 prender 41 preng- 13 th c. FGN:24, FNV:3 27 42 14 th c. IJ8:12, NOV:11 23 43 th 44 15 c. 0 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 45 of 51 Bulletin of Hispanic Studies

44 1 2 3 Table 3.3 4 5 6 Verb Century Texts and occurrences Totals 7 asir 8 asg- 13 th c. 0 9 14 th c. LBT:1 1 10 th 11 15 c. 0 12 doler 13 dolg- 13 th c. 0 14 ~duelg- 14 th c. IJ8:2 2 15 th 16 15 c. G&G:5, IMI:1 6 17 soler 18 solg-~suelg- 13 th c. For Peer Review 0 19 14 th c. ALX:1, IJ8:1 2 20 th 21 15 c. BAE:1 1 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 46 of 51

45 1 2 3 Table 3.4a 4 5 6 Verb Century Texts and occurrences Totals 7 caer 8 caig- 13 th c. 0 9 14 th c. IJ8:7 7 10 th 11 15 c. C01:3, CBO:3, CIR:2, CIS:2, CMN:15, CUZ:3, DON:9, 48 12 GOR:1, IMI:8, PN5:1, SL3:1 13 cay- 13 th c. --- -- 14 14 th c. --- -- 15 th 16 15 c. BAE:6, C01:1, CBO:1, CMN:2, DON:1, ENC:2, GOR:11, 32 17 HER:1, IMI:1, MEL:1, PER:1, PN5:2, Y89:2 18 huir For Peer Review 19 huig- 14 th c. IJ8:7 7 20 15 th c. C01:1, CBO:1, CLV:1, DON:1, IMI:2, PER:2, PN5:2 10 21 34 22 huy- 14th c. IJ8:5 5 23 15th c. C01:1, CBO:4, DON:3, IMI:3, PN5:5 16 24 oír 25 oig- 13 th c. 0 26 14 th c. AC5:2, IJ8:10 12 27 th 28 15 c. BAE:5, C01:10, CBO:1, CLV:1, CMN:3, CUZ:3, DON:4, 36 29 PER:6, PN5:1, PRS:1, SL3:1 30 oy- 13 th c. -- 31 14 th c. -- 32 th 33 15 c. BAE:5, C01:3, CBO:4, DAN:1, DIV:1, ENC:2, G&G:4, 51 34 GOR:5, IMI:9, MEL:4, OLI:1, PN5:1, TR1:2, Y89:9 35 traer 36 traig- 13 th c. FGN:1 1 37 th 38 14 c. AC5:3, IJ8:18, LBG:1, LBT:2, LEO:3, MON:2, REY:2, 32 39 TRS:1 th 40 15 c. BAE:9, C01:11, CBO:7, CIR:4, CIS:3, CLV:1, CMN:21, 96 41 CUZ:3, DAN:3, DIV:1, DON:5, G&G:1, GOR:1, IMI:6, 42 OLI:4, PER:11, PN5:1, PRS:1, VAS:2, Y89:1 43 tray- 13 th c. --- -- 44 th 45 14 c. AC1:3, CD1:3, CD2:5, CID:6, IJ8:6, LBG:9, LBS:12, 73 46 LCB:4, LEO:1, MON:4, NOV:1, REY:4, SPC:6, TRS:9 47 15 th c. BAE:7, C01:2, CBO:1, CIR:1, CIS:1, ENC:20, GOR:51, 111 48 HER:2, MEL:10, PN5:4, PRS:1, SUM:3, Y89:8 49 50 51 34 52 Etymological forms of huir presented are only those documented in texts which displayed 53 54 analogical forms. Documentation of etymological forms was not pursued across the entire 55 56 57 corpus. 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 47 of 51 Bulletin of Hispanic Studies

46 1 2 3 Table 3.4b 4 5 6 Verb Century Texts and occurrences Totals 7 destruir 8 destruig- 14 th c. IJ8:9 9 9 th 10 15 c. 0 11 leer 12 leig- 14 th c. IJ8:1 1 13 15 th c. 0 14 raer 15 th 16 raig- 14 c. IJ8:4 4 th 17 15 c. CMN:1 1 18 (son)reír For Peer Review 19 sonrig- 14 th c. 0 20 th 21 15 c. BAE:1 1 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 48 of 51

47 1 2 3 References 4 5 6 Alvar, Manual, 1953. El dialecto aragonés (Madrid: Gredos). 7 8 Alvar, Manuel, and Bernard Pottier, 1983. Morfología histórica del español (Madrid: Gredos). 9 10 11 Badía Margarit, Antonio, 1951. Gramática histórica catalana (Barcelona: Noguer). 12 13 Corfis, Ivy A., and John O’Neill, eds., 1997. Early Celestina Electronic Texts and 14 15 Concordances , CD-ROM (Madison: Hispanic Seminary of Medieval Studies). 16 17 18 Corominas, Joan, and ForJosé A. Pascual, Peer 1980. Diccionario Review crítico etimológico castellano e 19 20 hispánico , vol. 1 (Madrid: Gredos). 21 22 Correas, Gonzalo, 1625 [1954]. Arte de la lengua española castellana , ed. Emilio Alarcos García 23 24 25 (Madrid: CSIC). 26 27 Davies, Mark, 2002-. Corpus del Español (100 million words, 1200s-1900s) , available online at 28 29 http://www.corpusdelespanol.org . 30 31 32 Dworkin, Steven N., 1983. ‘The Fragmentation of the Latin Verb TOLLERE in Hispano- 33 34 (Including Luso-) Romance’, Romance Philology , 37: 165-173. 35 36 37 Elson, Mark J., 1988. ‘The Synchronic Status and the Evolution of the g in Spanish vengo , salgo , 38 39 etc. Revisited’, Hispania , 71: 392-400. 40 41 Elvira, Javier, 1998. El cambio analógico (Madrid: Gredos). 42 43 44 Faulhaber, Charles B., et al., 1984. Bibliography of Old Spanish Texts , 3rd ed. (Madison: 45 46 Hispanic Seminary of Medieval Studies). 47 48 Grandgent, C. H., 1905. An Outline of the Phonology and Morphology of Old Provençal , rev. ed. 49 50 51 (Boston, New York, and Chicago: D. C. Heath and Co.). 52 53 _____, 1962. An Introduction to Vulgar Latin (New York: Hafner). 54 55 Hanssen, Federico, 1913. Gramática histórica de la lengua castellana (Halle: Niemeyer). 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 49 of 51 Bulletin of Hispanic Studies

48 1 2 3 Herrera, Ma Teresa, and M a Estela González de Fauve, eds., 1997. Textos y concordancias 4 5 6 electrónicos del Corpus Médico Español , CD-ROM (Madison: Hispanic Seminary of 7 8 Medieval Studies). 9 10 11 Hualde, José Ignacio, 2005. The Sounds of Spanish (Cambridge: Cambridge University Press). 12 13 Imhoff, Brian, 2000. ‘Dialect Contact and Historical Linguistic Variation: The Old Spanish “-IE 14 15 Imperfect”’, Hispanic Review , 68: 381-396. 16 17 18 Jonxis-Henkemans, WilhelminaFor L.,Peer and Jerry R. Craddock,Review eds., 1992. Text and Concordance of 19 20 Fuero Juzgo . Hispanic Society of America MS. B2567 (Madison: Hispanic Seminary of 21 22 Medieval Studies). 23 24 25 Juilland, Alphonse, and E. Chang-Rodríguez, 1964. Frequency Dictionary of Spanish Words 26 27 (The Hague: Mouton & Co.). 28 29 Kasten, Lloyd , John Nitti, and Wilhelmina Jonxis-Henkemans, eds., 1997. The Electronic 30 31 32 Texts and Concordances of the Prose Works of Alfonso X, El Sabio , CD-ROM (Madison: 33 34 Hispanic Seminary of Medieval Studies). 35 36 37 Kauffeld, Cynthia, 2007. ‘Old Spanish -ir/-ecer Verb Variation: Tracing the Extension of -ec- 38 39 through the Lexicon’, Bulletin of Spanish Studies , 84: 965-987. 40 41 Klausenburger, Jürgen, 1984. ‘The Morphology of the Velar Insert in Romance Verbs’, in 42 43 44 Romanitas: Studies in Romance Linguistics , ed. Ernst Pulgram (Ann Arbor: University of 45 46 Michigan), pp. 132-151. 47 48 Kuryłowicz, J., 1968. ‘The Notion of Morpho(pho)neme’, in Directions for Historical 49 50 51 Linguistics , ed. Winfred P. Lehman, Yakov Malkiel (Austin: University of Texas Press), 52 53 pp. 65-81. 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 50 of 51

49 1 2 3 Lenfest, Donald E., 1978. ‘An Explanation of the /g/ in “tengo”, “vengo”, “pongo”, “salgo”, and 4 5 6 “valgo”’, Hispania , 61: 894-904. 7 8 _____, 1993. ‘Tengo-Vengo , an Update’, Hispania , 76: 634-644. 9 10 11 Lloyd, Paul M., 1987. From Latin to Spanish , vol. I: Historical Phonology and Morphology of 12 13 the Spanish Language (Philadelphia: American Philosophical Society). 14 15 Mackenzie, David, 1997. A Manual of Manuscript Transcription for the Dictionary of the Old 16 17 18 Spanish LanguageFor, ed. Ray Peer Harris-Northall, Review 5th ed. (Madison: Hispanic Seminary of 19 20 Medieval Studies). 21 22 Maiden, Martin, 2001. ‘Di nuovo sulle alternanze “velari” nel verbo italiano e spagnolo’, 23 24 25 Cuadernos de Filología Italiana , 8: 39-61. 26 27 Malkiel, Yakov, 1974. ‘New Problems in Romance Interfixation (I): The Velar Insert in the 28 29 Present Tense (with an Excursus on -zer/-zir Verbs)’, Romance Philology , 27: 304-55. 30 31 32 McMahon, April M. S., 1994. Understanding Language Change (Cambridge: Cambridge 33 34 University Press). 35 36 37 Menéndez Pidal, Ramón, 1964. Orígenes del español , 5th ed. (Madrid: Espasa-Calpe). 38 39 _____, 1968. Manual de gramática histórica española , 13th ed. (Madrid: Espasa-Calpe). 40 41 Meyer-Lübke, W., 1923. Grammaire des langues romanes , vol. 2: Morphologie , trans. Auguste 42 43 44 Doutrepont, Georges Doutrepont (New York: Stechert). 45 46 Moll, Francisco de B., 1952. Gramática histórica catalana (Madrid: Gredos). 47 48 Mondéjar, José, 1995. ‘Los presentes de subjuntivo anómalos en -g-, -y-, -ig-. Intento de 49 50 51 explicación’, in Estudis de lingüística i filologia oferts a Antoni M. Badia i Margarit 52 53 (Barcelona: Departament de Filologia Catalana, Universitat de Barcelona/Publicacions de 54 55 l'Abadia de Montserrat), pp. 11-23. 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Page 51 of 51 Bulletin of Hispanic Studies

50 1 2 3 Nebrija, Antonio de, 1492 [1980]. Gramática de la lengua castellana , ed. Antonio Quilis 4 5 6 (Madrid: Nacional). 7 8 Nitti, John, and Lloyd Kasten, eds., 1997. The Electronic Texts and Concordances of Medieval 9 10 11 Navarro-Aragonese Manuscripts , CD-ROM (Madison: Hispanic Seminary of Medieval 12 13 Studies). 14 15 O’Neill, John, ed., 1999. Electronic Texts and Concordances of the Madison Corpus of 16 17 18 Early Spanish ForManuscripts Peer and Printings ,Review CD-ROM (Madison: Hispanic Seminary of 19 20 Medieval Studies). 21 22 Penny, Ralph, 1991. A History of the Spanish Language (Cambridge: Cambridge University 23 24 25 Press). 26 27 _____, 2002. A History of the Spanish Language , 2nd ed. (Cambridge: Cambridge University 28 29 Press). 30 31 32 Pérez Saldanya, Manuel, 1995. ‘Un problema de morfologia verbal romànica: les formes verbals 33 34 velaritzades’, in Actas del I Congreso de la Lengua Española en América y España , ed. 35 36 37 María Teresa Echenique, Milagros Aleza, María José Martínez (Valencia: Universitat de 38 39 València), pp. 409-420. 40 41 Peset, Mariano, Juan Gutiérrez Cuadrado, and Josep Trenchs Odena, 1979. Fuero de Úbeda 42 43 44 (Valencia: Universidad de Valencia, Secretariado de Publicaciones). 45 46 Pountain, Christopher J., 2001. A History of the Spanish Language through Texts (London: 47 48 Routledge). 49 50 51 Real Academia Española: Banco de datos (CORDE) [en línea]. Corpus diacrónico del español , 52 53 http://www.rae.es . 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN Bulletin of Hispanic Studies Page 52 of 51

51 1 2 3 Ridruejo, Emilio, 1998. ‘La inserción de -g- en el presente de caigo , oigo , traigo ’, in Actas del 4 5 6 IV Congreso Internacional de Historia de la Lengua; La Rioja, 1-5 de abril de 1997 , ed. 7 8 Claudio García Turza, Fabián González Bachiller, Javier Mangado Martínez (Logroño: 9 10 11 Universidad de la Rioja, Servicio de Publicaciones), pp. 725-734. 12 e 13 Staaff, Erik, 1907. Étude sur l’ancien dialecte léonais d’après des chartes du XIII siècle 14 15 (Uppsala: Almquist & Wiksell). 16 17 18 Sturtevant, E. H., 1961.For Linguistic Peer Change: An IntroductionReview to the Historical Study of Language 19 20 (Chicago: University of Chicago Press). 21 22 Ureña y Smenjaud, Rafael de, 1935. Fuero de Cuenca (formas primitiva y sistemática: texto 23 24 25 latino, texto castellano y adaptación del ‘Fuero de Iznatoraf’) (Madrid: Academia de la 26 27 Historia). 28 29 Urrutia Cárdenas, Hernán and Manuela Álvarez Álvarez, 1983. Esquema de morfosintaxis 30 31 32 histórica del español (Bilbao: Universidad de Deusto). 33 34 Wang, William S-Y., 1969. ‘Competing Changes as a Cause of Residue’, Language , 45: 9-25. 35 36 37 Williams, Edwin B., 1968. From Latin to Portuguese: Historical Phonology and Morphology of 38 39 the Portuguese Language , 2nd ed. (Philadelphia: University of Pennsylvania Press). 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 The University of Liverpool, Liverpool L69 7ZN