Show Less
Open access

En torno a ‘haber’

Construcciones, usos y variación desde el latín hasta la actualidad


Edited By Carlota de Benito Moreno and Álvaro S. Octavio de Toledo y Huerta

Este volumen constituye la primera monografía que aborda el haz entero de construcciones en las que, a lo largo de la historia, ha participado el auxiliar más conocido del español, HABERE > haber. Catorce especialistas de universidades europeas y americanas trazan, desde muy diversas ópticas teóricas (lingüística de área, gramática formal, gramática de construcciones, sociolingüística, dialectología o gramaticalización), una perspectiva de conjunto, exhaustiva en los detalles y novedosa en sus hipótesis, sobre los distintos caminos que históricamente ha emprendido el verbo ‘haber’ y hoy caracterizan su comportamiento en español (y otras lenguas románicas aquí abordadas: catalán, francés, portugués...), adentrándose igualmente en aspectos generales de la variación y el cambio morfosintácticos.

Show Summary Details
Open access

Perfects in Contact on the Iberian Peninsula: Ibero-Romance, Arabic, and the Charlemagne Sprachbund

| 281 →

Bridget Drinka

The University of Texas at San Antonio

Perfects in Contact on the Iberian Peninsula: Ibero-Romance, Arabic, and the Charlemagne Sprachbund

In recent studies on the distribution of the periphrastic perfect (Eng. I have eaten, Span. he comido) on the Iberian Peninsula and in Europe in general, a new emphasis on the interactive role of geographical and sociohistorical factors has emerged. This scholarly trend is perhaps best exemplified by the comprehensive dissertation of Javier Rodríguez Molina (2010), who assembles chronologically- and geographically-stratified data, both literary and non-literary, in order to trace the diffusion of the category across space and time. What Rodríguez Molina discovers is that the perfects of the eastern regions of the Iberian Peninsula were the first to undergo grammaticalization and that Frankish influence from across the Pyrenees may have played a significant role in this development. In the present article, I provide support for and expand upon Rodríguez Molina’s findings, but take a step beyond his claims in asserting not only that the role of the Franks was essential in providing a model for the development of the periphrastic perfects in the east, but also that Arabic may have played a role in the constrained use of the perfects in the western region.

In what follows, I will proceed to analyze the effects of these two major external influences, Carolingian Latin and Arabic, on the Romance languages of Europe and, more precisely, on the varieties of the Iberian Peninsula. After a brief look at the claims of Rodríguez Molina, our attention will turn to an examination of the influence of the “Charlemagne Sprachbund” on the perfects of Europe, followed by an investigation of the Iberian varieties from west to east, including a consideration of the possible role of Arabic in influencing the outcome of the perfects in western Spain and Portugal.

1.1 The Role of the Franks: Rodríguez Molina (2010)

In his remarkably thorough analysis of the grammaticalization of the periphrastic perfects in Old Spanish, Rodríguez Molina (2010: 1219–25) finds that the HAVE perfects of the eastern region of the Iberian Peninsula, especially the varieties of ← 281 | 282 → Aragón and Navarra, were more frequent and more grammaticalized at an early time than those of any other region of the peninsula. Recognizing the essential role of contact and koineization in the history of Spanish (Penny 2000; Tuten 2003), Rodríguez Molina (2010: 1224–25) proposes that one of the explanations for this early grammaticalization in eastern varieties was the presence of a large Frankish population in Jaca, Huesca, Pamplona, and Estella (Zone A on Map 1) who took part in the “repoblación” of the territory and who garnered considerable prestige. Only later, when the Kingdom of Castile was joined to Navarra in the 11th century, did many Aragonese settlers move into eastern Castile, fostering the growth of the perfect there (Zone B). In the western half of the peninsula (Zones C and D), on the other hand, the influence of the Franks was significantly less, and the growth of the periphrastic perfect was negligible.

Map 1: The grammaticalization of haber + ptcp: isoglosses and dialectal areas (after Rodríguez Molina 2010: 1219)


These trends are clearly recognizable in the notarial records of the various regions: the frequency of the HAVE + PP construction in the eastern varieties exceeds that of Castile and far exceeds that of León throughout the medieval period (Figure 1). ← 282 | 283 →

Figure 1: Frequency of have + pp in notarial documents (after Rodríguez Molina 2010: 1214)


These data support Rodríguez Molina’s claim that the innovation began in the east, due largely to Frankish influence, and spread westward from there.

1.2 Core vs. Periphery in the Charlemagne Sprachbund

To what extent are the perfects of trans-Pyrenean Romance languages distinct from those of the Ibero-Romance varieties, and what could have caused the former to have influenced the latter? We begin our examination of possible Frankish influence on Iberian perfects with a look at the general distribution of perfects and other morphosyntactic categories across the map of western Europe.

Johan van der Auwera (1998b: 824), noting that French, German, and Dutch share a number of similarly-constructed grammatical patterns, and that these varieties are all located within the area originally ruled by Charlemagne, coined the term Charlemagne Sprachbund to recognize the existence of a western European linguistic area. Among the features he considered most diagnostic are the semantic shift of anterior > preterite, the use of future perfects, and the presence of supercompound pluperfects (van der Auwera 1998b: 833 note 13). When key morphosyntactic features like these are tallied and plotted on an isopleth map, a distribution such as that seen in Map 2 is found. ← 283 | 284 →

Map 2: Charlemagne Sprachbund isopleth map for tense-aspect (van der Auwera 1998b: 826, based on Thieroff 2000)


This mapping of the salient temporal-aspectual features of the languages of western Europe points to French and German as core members, and other Romance and Germanic languages as more peripheral (van der Auwera 1998b: 826), based on Thieroff (2000: 265–305).

The periphrastic perfect turns out to provide remarkably clear evidence not only that anteriors became pasts in the “core” languages—a development which, in fact, occurred well after the time of Charlemagne—but even more importantly, that the BE/HAVE dichotomy came to be especially developed and maintained in those languages within Charlemagne’s realm, and less so in those that lay without. An examination of the present-day distribution of the periphrastic perfects on the map of western Europe (Map 3)1 reveals both geographical and chronological components, showing the effect of centrality vs. peripherality as well as the chronological layering of Stages I–III. ← 284 | 285 →

Map 3: The present-day distribution of have and be perfects in western Europe (adapted and revised from Dahl / Velupillai 2013)

STAGE I. Present-day area where HAVE perfects occur[HAVE]+[HAVE/BE]+[PERF > PRET]
STAGE II. Within the HAVE area, division of labor between HAVE perfects and BE perfects[HAVE / BE]+[PERF > PRET]
STAGE III. Within the HAVE / BE area, anterior meaning of perfect has developed into past or perfective[PERF > PRET]
Peripheral area with less typical HAVE perfects[P] ← 285 | 286 →

Several distinctive areal patterns are recognizable on this map: we note that the existence of a HAVE perfect is widespread in western Europe (I), with a smaller, “core” area employing both the HAVE and BE auxiliaries especially in French, German, and northern Italian (II), and an even smaller “nuclear” area representing the semantic shift of anterior > preterite extending across several contiguous varieties, including northern French, southern German, and northern Italian (III). As mentioned above, these innovations are chronologically stratified, with Stage I occurring first, II next, and III most recently. They can, then, be conceived of three-dimensionally, stacked one atop the other.

When we now compare this map to the boundaries of the Carolingian Empire at the time of Charlemagne’s death in 814, we find an even more remarkable coalescence:

Map 4: Revised WALS map of Perfect compared to Charlemagne’s Empire (814 ad)


← 286 | 287 →

The boundaries of the Carolingian realm, marked in bold, coincide precisely with the distribution of HAVE/BE auxiliation, that is, Stage II. The only notable outliers are Danish and Breton, which adopted the HAVE/BE contrast through heavy contact with German and French, respectively.

How can one account for the unified nature of this linguistic area? Carolingian territories were clearly linked by strong ties: Charlemagne established a network of officials, missi dominici, who had the power to act on his behalf and who helped develop the efficient system of communication across the realm. He also communicated by means of capitularies and by calling assemblies in various locales. The fact that scribal tradition was refined and well-diffused across the Empire but is not traceable to any one scriptorium points to the widespread and fully-established nature of this tradition, and of Carolingian influence across the entire realm (McKitterick 2008: 370–80).

In order to build a stronger Christian society, Charlemagne reorganized the Church, continuing his father Pippin’s practice of replacing the “Gallican” rite of Merovingian Gaul with the Roman rite, and establishing monastic and episcopal schools for educating clerics in Latin. Latin was recognized as the official language of the realm, and was adopted as a symbol of Frankish identity and authority, with its historical clout and connection to the Roman Empire, and as a conduit of Christian ideals and traditions.

Granted that the political and social structures of the Carolingian realm fostered unified scribal tradition, one must still wonder why the division of “core” and “periphery” would have come to express itself in the selection of auxiliaries. In other words, why should the innovative increase in periphrastic perfects correspond so well with the territory of Carolingian dominance, and what role does Latin play in this distribution? We will examine the evidence for HAVE and BE constructions in turn.

1.3 HAVE perfects in Carolingian Latin

The HAVE perfect has a long and complex history in classical Latin, but it is in the early medieval documents, especially those produced in Gaul, where a noteworthy increase in productivity is to be found, above all in the juridical formulas of Late Latin:

(1) Formula of Lindenbrogius p. 280, 18

 Utrumille homohoc homicidium perpetratum haberet
Whether that man this homicide perpetrated HAVE.IMPRF.SUBJ.3SG
‘Whether that man had committed this murder’ ← 287 | 288 →

Other similar formulas appear in Gaul from the 6th c. on: habeo promissum, licitum, cessum, delegatum, iniunctum, etc. (‘have promised, permitted, yielded, delegated, enjoined’), as well as such frequent collocations as concessum atque indultum habeo ‘have conceded and permitted (lit. indulged)’. Most noteworthy in this innovative expansion is the more frequent use of the HAVE perfects in the Carolingian capitularies (Thielmann 1885: 545–47):

(2) Capitulary of Charlemagne 146, 32 and 146, 34

          sicut domnus imperator mandatum habet
 thuslordemperororderedHAVE. PRS.3SG
 ‘thus has the Lord Emperor ordered

(3) Capitulary of Charlemagne 175, 25; 183, 33

          secundum quod iudicatum habemus
 ‘in conformity with what we have ruled

Particularly indicative of the greater productivity of the HAVE construction in the Carolingian capitularies is the overextension of its use: parallel to the more normative use of the synthetic pluperfect subjunctive (4), a “less logical” use with the HAVE perfect also appears (5) (Thielmann 1885: 546):

(4) Capitularies of Charlemagne 163,15 [capit. 6]

     ‘how he had relinquished the secular world’

(5) Capitularies of Charlemagne 163,7 [capit. 5]

      Si illesaeculumdimissum habeat
     ifthatworldrenounced HAVE.PRS.SUBJ.3SG
     ‘if he has renounced the secular world’
     (examples from Thielmann 1885: 545–6)

This expansion of the HAVE perfect in Carolingian legal documents comes to be reflected in legal and literary corollaries in the Romance and Germanic languages. The fact that the other major context for expansion of the HAVE perfects—the recording of direct speech—reflects colloquial usage of the construction also points to an incipient trend in the vernacular languages which will undergo significant expansion in the following centuries (Brinkmann 1931: 28).

1.4 Deponents and BE perfects in Carolingian Latin

Not only did the HAVE perfects flourish in particular contexts in Carolingian Latin, but the BE perfects also grew in prominence. As shown in (6) and (7), the ← 288 | 289 → passives and deponents2 are formally identical, forming periphrastics with a BE auxiliary in the perfect and pluperfect:

(6) Latin Perfect Passives

Presentlaudor‘I am praised’
Perfectlaudātus sum‘I was /have been praised’
Imperfectlaudābar‘I was being praised’
Pluperfectlaudātus eram‘I had been praised’

(7) Latin Perfect Deponents

Presentsequor‘I follow’
Perfectsecūtus sum‘I followed / have followed’
Imperfectsequēbar‘I was following’
Pluperfectsecūtus eram‘I had followed’

As I argue elsewhere (Drinka 2013; forthcoming), it was the growing productivity of the deponents and passives in Late Latin which led to the establishment of the BE perfect across the Carolingian realm, for just as the synthetic passives came to be replaced by periphrastic BE forms, the expanding array of deponents and “intrinsic passives” were likewise replaced by the BE perfects. Flobert argues that Latin and the vernacular shared the responsibility for the development of the periphrastics by influencing each other:

Le réajustement de l’écrit sur le parlé—et vice-versa—est continuel […]; l’évolution rapide du latin qui se traduit, après l’éclatement de la Romania, dans les “fautes” du latin mérovingien et carolingien provient incontestablement de l’action du modèle parlé.3 (Flobert 1975: 589)

While the synthetic deponents were used by some Late Latin writers to add an air of refinement and literary finesse to their writing (Norberg 1943: 154–5), the periphrastic forms eventually prevailed, both in the Late Latin of Charlemagne’s time and in the early vernaculars of France and Italy. It is precisely this fact which, I would claim, is reflected on Map 4. The effects of the Charlemagne Sprachbund persist to this day in the distribution of the BE and HAVE dichotomy.4 ← 289 | 290 →

1.5 Linguistic and stylistic evidence: The Annales Regni Francorum

A brief look at a quintessentially Carolingian document, the Annales Regni Francorum (ARF), or Royal Frankish Annals (741–829), will demonstrate the role of the perfect across time. While the early annals show greater variability, the later annals illustrate increasing complexity across time, as well as a growing use both of the deponents and of the BE perfect. The annals which represent the most stylistic and morphosyntactic unity in the entire ARF are those from 820–29, presumably written by the abbot of Saint-Denis, Hilduin.

Several examples from the ARF will illustrate these trends and correlations:

(8) 741 (early)

Carolus maior domus defunctus est.5

‘Charles [Martel], mayor of the palace, died.’

(9) 787 (mid)

Et cum venisset ad hoc locum, quod omnia explanasset de parte Tassilonis, sicut actum erat, tunc prespiciens idem rex, ut missos mitteret, et iussit Tassiloni, ut omnia adimpleret secundum iussionem apostolici, vel sicut iustitia erat: eo quod sub iureiurando promissum habebat, ut in omnibus oboediens et fidelis fuisset domno rege Carolo et filiis eius vel Francis et veniret ad eius praesentiam; quod rennuit et venire contempsit.

‘And when he had come to that place, he explained the affair with Tassilo, just as it had occurred, then the king decided to send emissaries and ordered Tassilo to do everything according to the pope’s instruction and the demands of justice, since he had promised under oath that he would be obedient and loyal in everything to the Lord King Charles, his sons, and the Franks, and that he would appear before him. But Tassilo rejected this and refused to come.’

(10) 820 (late)

In eo conventu Bera comes Barcinonae, qui iam diu fraudis et infidelitatis a vicinis suis insimulabatur, cum accusatore suo equestri pugna confligere conatus vincitur. Cumque ut reus maiestatis capitali sententia damnaretur, parsum est ei misericordia imperatoris, et Ratumagum exilio deportatus est.

‘At this assembly Count Bera of Barcelona, who for a long time had been accused by his neighbors of bad faith and treason, tried to contend with his accuser in combat ← 290 | 291 → on horseback but was defeated. He was first condemned to death for lèse majesté but then pardoned by the mercy of the emperor and taken away into exile to Rouen.’

As would be predicted for a narrative text like this, simple synthetic preterites (Latin perfects) predominate, but what these examples also illustrate is a growing tendency for scribes to use synthetic deponents and passives, BE periphrastics, and more complex morphosyntax (subjunctives, pluperfects, etc.), especially in the late period. These tendencies turn out to be statistically significant (Drinka forthcoming).

Some mention must be made of the fact that very few examples of the HAVE perfect appear in the ARF: a total of 5 HAVE pluperfects appear, three in the middle period, in 787, 788, and 791 (see (9) above), and two more in the later period, in 822 and 829—both, remarkably, with the PPP dispositum.6 HAVE perfects, then, were not excluded, but they represent a miniscule percentage of the periphrastic perfects and passives to be found in the ARF. Why should HAVE perfects be less well-represented in the annals of the Frankish kings than in their capitularies? It is clear that genre plays a significant role in this distribution: while both the annals and the capitularies use deponents and BE periphrastics extensively, the annals utilize a narrative style which was evidently less conducive to the production of HAVE perfects than the legal language of the capitularies was.

1.6 Summary of the role of the Charlemagne Sprachbund

Several conclusions can be drawn concerning the role of the Charlemagne Sprachbund in establishing the configuration of the periphrastic perfect in western Europe:

1. As a result of the political and social ties established during the time of Charlemagne, innovative strengthening of the dyadic relationship between HAVE and BE tended to occur in the core area, with BE periphrastics undergoing significant growth; in the peripheral areas, this growth did not occur. The ← 291 | 292 → borders of the Carolingian realm correspond closely to the boundaries of the HAVE / BE core.

2. The role of Carolingian Latin in establishing this innovation, with its tendency to use deponents and passives in ever-expanding contexts, is clear. Scribes relied on classical models, constructing increasingly complex morphological patterns, including a profusion of deponents. The vernacular correspondents of these deponents and synthetic passives were periphrastic BE perfects and passives. This style of writing Latin was diffused across the realm through a strong network of missi dominici, capitularies, and other officials. Outside of the Carolingian territory, older conventions of Latin tended to persist, such as the Visigothic tradition on the Iberian Peninsula. Because the innovative reinvigoration and expansion of BE perfects did not reach these areas, only the eastern varieties developed this dichotomy.

3. The Annales Regni Francorum, or Royal Frankish Annals, recording the exploits and accomplishments of the Carolingian kings between 741 and 829 AD, provides crucial evidence that the BE periphrastics followed a similar and related trajectory to the deponents: late in the period, the use of the BE periphrastic perfects and passives was shown to be highly correlated to the use of the synthetic deponents and passives. It is ultimately the cross-influence of written and spoken varieties of Latin and Romance vernaculars which, I claim, led to the increase in BE perfects to be witnessed in the Carolingian realm.

2. The core and peripheral features of the varieties of the Iberian Peninsula

Like the varieties of the Italian Peninsula, those of the Iberian Peninsula show several gradations of peripheral and core qualities, depending on their geographical location and the sociohistorical pressures that they have experienced. We will begin this analysis with a brief look at the history of Medieval Spain, and will examine the role of early Visigothic and Andalusian contacts as influential and conservatizing forces. We will then proceed to examine varieties from west to east, beginning with a special focus on Portuguese as an extreme example of a peripheral variety, and the nearby varieties of Galician, Leonese, and Asturian, followed by Castilian—a clear but less radical example of peripherality—Aragonese, and Catalan. What we will find as we move from west to east is ever more affinity to the patterns found in French and northern Italian, that is, to the perfects of the Charlemagne Sprachbund. ← 292 | 293 →

2.1 Spanish and Portuguese as peripheral languages: conservative tendencies

An examination of the history of early medieval Spain will give some indication as to why Castilian Spanish and especially Portuguese remained peripheral languages from a European perspective. Visigothic rule (6th–7th centuries) followed by Muslim control of three-quarters of the peninsula (8th–11th centuries) (Map 5) precluded Spanish adoption of innovations from north of the Pyrenees until the late 11th century (Wright 1982: 150–51, 165). The dearth of HAVE perfects in the early medieval Latin of the Iberian peninsula, as compared to the relative abundance in the Latin of Gaul, provides clear evidence of this isolation.7

Map 5: Christian & Muslim Territories and the Expansion of Castile (Lapesa 1968: 136)


The Visigothic rite continued to be practiced by the Mozarabs, that is, the Christians of Al-Andalus, and by those Christian exiles from Toledo who had fled from the southern Muslim territories to the northwest (Beale-Rosano-Rivaya 2006). This retention of Visigothic ritual and beliefs met with disapproval from the Carolingians, who judged these beliefs to be heretical. The Benedictine monastic ← 293 | 294 → tradition, pervasive in the Carolingian realm, did not spread to Spain until much later, and it is therefore likely that Latin as written and spoken north of the Pyrenees was not used in Spain before the adoption of the Roman Liturgy in 1080. It was the adoption of this liturgy, imposed by Pope Gregory VII, which was instrumental in the decision of the county of Portugal to secede from Leonese rule (Wright 1982: 210). One could, in some sense, say that Portugal seceded not only from León but from the Charlemagne Sprachbund itself in choosing not to accommodate to the traditions and rituals, and, hence, the linguistic influence which came down from the north.

What effect did these political and historical events have on the development of the perfects? One eventual consequence of the political split between Spain and Portugal appears in the separate trajectories that Spanish and Portuguese auxiliaries HABERE and TENERE took in the centuries following this division. When the auxiliary data assembled by Harre (1991) is reconfigured as percentages and arranged chronologically as in Table 1 and Figure 2, we note that speakers of medieval Spanish and Portuguese were following a very similar trajectory in their auxiliary use at an early time: both used both auxiliaries at fairly similar rates, and both used HABERE more frequently. However, in the 15th century, the two varieties underwent a decided split. Castilian began to follow an upward trend in using haber more frequently than tener, while Portuguese took the opposite tack, and opted for increased use of ter at the expense of haver.

Table 1: Comparison of rate of Spanish haber use to Portuguese haver use over time (based on Harre 1991)8


← 294 | 295 →

Figure 2: Comparison of habere and tenere in Spanish and Portuguese over time (based on Harre 1991)


Why should these two varieties have taken such different paths at this particular point in time? What I will claim is that Portuguese continued to rely on more vernacular tendencies that had developed on the Iberian Peninsula, influenced to some extent by contact with Arabic, while Castilian moved instead at least partially in the direction of trans-Pyrenean Europe, reinforcing the use of haber as an auxiliary as the Reconquista progressed. It will be the aim of the following sections to examine the evidence for these claims, to sort out the geographical and chronological distributions of the auxiliaries, and to determine the extent to which political allegiance and other sociohistorical factors were responsible for this outcome.

2.2 The influence of Arabic

Scholars tend to downplay the role of Arabic in the history of Spanish and Portuguese, viewing this variety as a source of lexical borrowing, but not as a model for structural repatterning.9 What I will argue is that the seven-century-long dominance of Arabic on the Iberian Peninsula could not have vanished ← 295 | 296 → without leaving some trace of structural influence, and that the periphrastic perfect may well provide evidence of that influence, specifically in the tendency of the perfect to express durativity or iterativity in a number of Romance varieties. As will be explored in the following pages, the geographical distribution of this semantic feature coincides remarkably well with the expanse of Muslim control in the Mediterranean: on the Iberian Peninsula, durativity and iterativity appear pervasively in the perfects of Portuguese and the nearby Spanish varieties; the feature is also found in diasporic Judeo-Spanish and across many varieties of Latin American Spanish, reflecting the earlier variable presence of this feature in Castilian; finally, it is also found in Sicilian, where Arabic rule existed from 827–1091.

2.2.1 Historical background of Al-Andalusia. As illustrated in Map 5, the Emirate of Córdoba, the Almohad Caliphate, and several independent Islamic states occupied the southern three-quarters of the peninsula in the 10th–12th centuries. While the early Umayyad kings ruled the sparse population according to harsh, conservative traditions and strict Islamic law, rulers of tenth-century Al-Andalus such as Caliph al-akam brought in scientific and philosophical learning from the East, including many books translated from Greek, and fostered the study of medicine and science, making Córdoba a major center both of secular and Islamic learning (Hourani 1972: 99). Muslim control of much of the Iberian Peninsula until the late eleventh century led to bilingualism and diglossia in the early years, and language shift to Arabic, in many cases, in later years (Beale-Rosano-Rivaya 2006: 66). Glick (1979: 177) goes so far as to claim that at the time of the conquest of Toledo in 1085 by northern Christian forces, the entire Christian population was monolingual Arabic-speaking. Wasserstein (1991: 12–15) and Wright (2002: 158–74) refine this view by noting that widespread ladino-Arabic bilingualism existed on the spoken level, but that, from the mid-ninth century on, where literacy existed among the Mozarabic Christians, the Muwallad converts to Islam, and the Jews of Al-Andalus, it existed in Arabic. Considerable evidence points to the widespread use of Arabic among Mozarabic Christians: Bishop Marino of Catalonia, for example, conducted Mass in Arabic for the Mozarabic congregation there; expressions from the Qur’ān were used to introduce the Christian gospels (“In the name of God, the Compassionate, the Merciful”); juridical texts found in the archives of the Cathedral of Toledo and in the monasteries and churches were still being produced in Arabic as late as 1391 (Beale-Rosano-Rivaya 2006: 112–29). Some martyrs of Córdoba are said to have pronounced their blasphemies against Muh̟ammad or Islam in Arabic, and one of the staunch Christian leaders in that city in the mid-ninth century, ← 296 | 297 → Albaro (Alvarus),10 complained that the Latins did not pay attention to their own tongue, but could produce elaborate displays of terms in Arabic (Wasserstein 1991: 3).11 It is clear that the Mozarabic community was thoroughly Arabized, and that the Reconquista of Toledo in 1085 did not immediately impact that well-entrenched tradition.

2.2.2 The perfects of Arabic. In order to assess the potential influence of Arabic on the development of the perfects on the Iberian Peninsula, we will briefly examine the construction of the perfects in Arabic. It is noteworthy that Standard Arabic12 does not have a designated perfect category, but rather conveys anterior meaning by means of the Perfective with adverbial modification (Amman 2002: 332; Fassi Fehri 2003: 71; 92):13

(11)Standard Arabic
        katab-a       r-risaalat-a       l-aan-a
        ‘he has written the letter now’
(12)maryam-u ntađ̟ar-atzaynab-amunđu saacatayni
Miryamwait.PRV-3SG.FEMZaynabsincetwo hours
‘Miryam has waited for Zaynab for two hours’
(examples from Fassi Fehri 2003: 78; 88) 

Comrie (1976: 81) claims that Classical Arabic had a Perfect category formed with the particle qad ‘already’ plus the Perfective: ← 297 | 298 →

(13)Classical Arabic
        qad kataba
        ‘he has written’

The presumed etymology of qad as deriving from the active participle qāεid ‘sitting’ could account for the relation of past action to present state: ‘he is sitting having written’ (Cuvalay-Haak 1997: 163; 239).

Maltese provides convincing evidence for the accuracy of this etymology while also hinting at the potential durative application of the construction: the very productive Maltese auxiliary ɂa˚ad ‘accomplished’, literally ‘sit, stay,’ cognate with Classical / Standard Arabic qad, denotes durativity and can occur with most verb classes:

‘after they had threatened him many times’ (Vanhove 2001: 70)

This auxiliary also occurs in other Arabic dialects, especially in the Maghreb of northern Africa, conveying related semantic values such as inchoativity or continuativity (Vanhove 2001: 70–72).

Besides the anterior use of the synthetic Perfective, especially as reinforced by the particle qad, Standard Arabic can also express the anterior by means of an active participle deriving from the noun system. For example, the finite active verb yaktub ‘he writes’ has an active participle kātib, meaning ‘a person who writes / a writer’. When these participles are used as adjectives, they often refer to a present state bounded at its beginning or end by a perfective event. If the event precedes the present moment, it implies retrospection, with a possible interpretation as a resultative. If the event follows the present, it implies a prospective, or a future (Kinberg 1992: 312):

(15)Retrospective use:Qu’ran 36: 16 
‘Our Lord knows that we have been sent on a mission to you.’
(16)Prospective use:Qu’ran 40: 49
’innas-sā ‘atala-’ātiyatun
Certainly ART-hourASSERTIVE-coming_thing
‘The Hour is coming’
(examples from Kinberg 1992: 313, 316)

When the reference point occurs prior to the time of utterance, the past tense of the auxiliary BE is used with the participle: ← 298 | 299 →

(17)Retrospective use:Qu’ran 28: 8
’innafir’awnawa-hāmāna wa-ğunūdahumākānūh̬āt̟i’īna
Certainly Pharaohand-Haman and-hosts-theirwere-3PL sinners
‘Certainly Pharaoh and Haman and their hosts were sinners /had sinned’
(example from Kinberg 1992: 309)

Kinberg (1992: 313) notes, however, that the use of the “retrospective semi-imperfectives” is “quite marginal” in the Qu’ran, probably due to the fact that the above-mentioned qad + perfective form already existed in Classical Arabic. The construction persists in most Modern Arabic dialects, however, and is much more frequent there (Cuvalay-Haak 1997: 175), as illustrated in (18):

(18)Gulf Arabic
       ‘The child has eaten his lunch’
(example from Amman 2002: 333)

What is crucial to note is that both constructions which express the anterior in Arabic, the perfective, especially as reinforced by qad, and the active participle construction, show some potential or actual connection with the durative.

2.2.3 Possible influence on Romance perfects. Similarities between the Arabic constructions and the periphrastic perfects of western Europe have been noted by several scholars. Amman (2002: 332–3) draws a parallel between the Arabic participle constructions and the periphrastic perfects, but notes that the Arabic varieties use active participles, in contrast to the passive participles of Romance and Germanic.14 Fassi Fehri (2003: 72) points more specifically to the distribution of preterites and perfects in Portuguese as parallel to that of Standard Arabic. In Portuguese, as in Arabic, the synthetic preterite is often used to mark the anterior, and the periphrastic perfect (ter ‘hold, have’ + PP), like the active participle construction, is used for habitual situations alone:15 ← 299 | 300 →

       Agora jà temcomido
       Now Ihave eaten
       ‘Now I have taken on the habit of eating’

While these proposed connections are intriguing, one must also wonder if the qad construction represents an even closer parallel, especially as it appears in Maltese, with an auxiliary meaning ‘sit, stay’ and obligatory durative meaning, not unlike the ter perfect of Portuguese.

What other morphosyntactic effects might we expect to see if Arabic did, indeed, influence the varieties of Spanish and Portuguese which came in close contact with it? We would predict something like the following:

1) parallel morphological developments

2) diminished use of periphrastic perfects as anteriors, with a corresponding increased use of synthetic preterites for this purpose

3) intensification of aspectual distinction in the verb system

The first criterion has been discussed above. Evidence for the second criterion, the decreased use of perfects, is to be found in Mozarabic Latin: my examination of fifty pages of Albaro’s letters (collected in Gil 1973) yielded a number of examples of HAVE used as a possessive verb (potestatem habeo ‘I have power,’ responsum habeto ‘have a response (2/3S.FUT.IMP),’ habet emolumentum ‘has advantage’), several BE perfects/deponents (natus est ‘is born,’ locutus est, ‘has spoken,’ mutata est ‘has changed’) and periphrastic passives (factus est ‘was made,’ perfecti sumus ‘we were finished’), but no HAVE perfects at all. Gil (1971: 203) points to the loss of the synthetic deponents, and connects this trend to hypercorrection and to the “hiperurbanismos” of Mozarabic writing; this development stands in stark contrast to the strengthening of the deponent category which was occurring in the Carolingian tradition at the same time.

Evidence for the diminished use of perfects due to Arabic influence can also be found in traces of vernacular Romance preserved in the xarjah poetry, short poems constituting the final refrain (hence the name xarjah ‘final’) of five-stanza muwašša poems. These poems, probably composed between the 10th and the 12th centuries, were written especially in Arabic, using Arabic script, and sometimes contained Romance lexicon:

(20)ké faré yo o ké šerád de míbi? | h̟abíbi, | nón te tólgaš de míbi.
‘what shall I do or what will become of me? My dear, do not forsake me’
(example from Corriente 1991: 62) ← 300 | 301 →

The xarajāt are often linguistically hybrid,16 but are composed especially in Andalusian Arabic or Classical Arabic. Only a small number actually contain Romance material—42 of 600 muwaššah̟āt (Jones 1991: 89). The fact that the xarajāt are frequently written from a woman’s perspective may be tied to the tendency for women, peasants, and slaves to be the prime users of this “prestigeless” Romance variety (Corriente 1991: 61; 66). Important insight into the role of the xarjah is provided by the 12th-century Egyptian writer Ibn Sanā’ al-Mulk in his treatise Dār al-T̟irāz, in which he describes the language of the xarajāt as that of “hooligans”:

The kharja used to be in a foreign language, in fact, an unintelligible tongue. The kharja is the spice of the muwašša, its salt, its sugar, its musk and its ambrosia. […] He who composes a muwašša arranges [the kharja] first thing even before he deals with metre and rhyme.[…] He found the basis, he secured the tail and built upon it the head (from Haxen 1991: 41–2)

This special genre of Andalusian poetry points to the complex intertwining of linguistic, literary, and musical traditions of the time; extracting from this mix the strands of Romance lexicon and structure, we can grasp in some small way the nature of the Spanish vernacular.17 With regard specifically to participial and perfect use, Corriente (1997: 354, ftnote 44) notes that past participles are used only as adjectives or to express past actions in the xarajāt, and are not attested in periphrastic constructions. While aber and tener are both used as verbs of possession, neither is used as an auxiliary. Corriente explains the lack of periphrastic perfects in this vernacular poetry as “un empobrecimiento del sistema debido, sin duda, al adstrato árabe.”18 What we may be witnessing, then, is the above-mentioned dampening effect of Arabic on the development of the perfects of Spanish.

The third criterion suggested above which might point to Arabic influence is an intensification of aspectual distinction. While it is difficult to demonstrate a direct influence, I believe that several pieces of evidence point to this conclusion:

1. the distribution of haber/tener grew to be parallel to ser/estar—a distinction made solely on the Iberian Peninsula ← 301 | 302 →

2. the evidence from Old Castilian and Old Portuguese points to early variability between haber and tener as HAVE auxiliaries, with attendant lexically-determined aspectual distinctions; while Castilian underwent a growth in the use of the periphrastic perfects, specializing the two verbs for HAVE as a verb of possession (tener) and an auxiliary (haber), Portuguese, Galician, and other western varieties underwent a drastic diminution in the frequency of the perfects, and adopted ter as both the verb of possession and the auxiliary.

To summarize, what I claim is that the Portuguese distribution may more accurately reflect the Romance varieties’ uninterrupted development spoken in Al-Andalusia under the influence of Arabic aspectual distinction, as likewise witnessed in the remnants of modern-day Judeo-Spanish, and, apparently, in New World Spanish. In contrast, I claim that the Castilian distribution represents, to some degree, a marking of allegiance with the rest of Europe in its strengthening of haber as perfect auxiliary. As we shall see, Castilian does not participate in the innovations of the Carolingian realm as Catalan and, to some extent, Aragonese did, having given up its BE auxiliaries in favor of HAVE. However, it also does not align itself with the vernacular tendencies of Portuguese. Evidence for these claims will be presented below, with a re-examination of their plausibility provided at the end.

3. Portuguese

As mentioned earlier, Portuguese took a definitive step away from the Carolingian model in breaking off from León in 1080, rather than adopting the Roman Liturgy. While the effects of this separation are not immediately reflected in the verbal patterns of medieval Portuguese, the eventual outcome is a system quite unlike that found in the core area of Europe, a system incontestably to be characterized as “peripheral.” What we observe in Portuguese over the course of the centuries following the secession is the persistence of the diminished role of the periphrastic perfect as marker of the anterior in favor of the synthetic preterite, the shrinking of the range of auxiliaries from a broader array to just one auxiliary, ter ‘hold, have’, and the retention of the iterative or durative value of the construction, clearly fostered by the semantic value of the auxiliary.

In Old Portuguese, the auxiliaries haver, ter, and ser were all used with a PP to form a perfect, but haver was the preferred auxiliary, as well as the most frequent independent verb of possession; ter appeared more rarely (Paiva Boléo 1936: 109). For transitives, haver or ter was used; for intransitives and unaccusatives, haver, ter, or ser appeared; for reflexives, ter (as is the case in Modern Portuguese), or, very infrequently, haver was chosen: ← 302 | 303 →

 Old Portuguese auxiliary use
‘you have called
tertenho vystos e ouvydos muitos enxempros
‘I have seen and heard many examples’
(22)Intransitive:haveras dormido
‘you have slept
tertiinha ele ja andadas duas leguas
‘he had already gone (intr.) or completed (tr.) 2 miles’
(23)Unaccusative:sersomos chegados19
‘we have (lit. are) arrived
(examples from Huber 1933: 221; Mattoso Camara 1972: 145)

In the 15th century, ter began to overtake haver in frequency as an independent verb, and to appear with participles, but not yet forming a “verbal unity”; in the 16th century, it clearly became the preferred auxiliary (Paiva Boléo 1936: 111).20 In contrast to a number of other Romance varieties, the Modern Portuguese ter construction is not used if the event is situated in the past, even if it occurred in the recent past, or is experiential, hodiernal, or represents hot news. It is restricted to durative and non-telic situations which include the Reference Time (Squartini / Bertinetto 2000: 409–10).21 Simply put, the function of the Portuguese present perfect is “to account for a present state in terms of a past event” (Algeo 1976: 204). The synthetic preterite is used where many other European languages would use a periphrastic perfect ← 303 | 304 → (Sten 1973: 239).22

The durative quality of the Portuguese perfect, mentioned above, is arguably produced by the likewise durative semantic value of the auxiliary itself, and constitutes an additional aspect of exceptionality in contrast to perfects found in the core area of Europe.23 Paiva Boléo (1936: 6) quotes the following popular poem to illustrate the contrast between synthetic preterites and the durative or iterative status of the ter + PP construction:

(24) À sombra do lindo céu

eu jurei, tenho jurado

não ter mais24 outros amores

só a ti tenho amado.

‘In the shadow of the lovely sky / I swore, I have sworn / not to have any more other loves; / only you have I loved

He suggests that tenho jurado implies iterativity, that the lover has sworn many times, while tenho amado refers to the durative quality of this loving, which lasts until the present. These instances of ter + PP might better be viewed in a unified manner as “permansive” (Matoso Camara 1972: 146 ftnote) or “protracted” (Algeo 1976: 197), rather than being differentiated as durative or iterative, since both the love and the effect of the oath continue to exist from the time of their first manifestation into the present.

In sum, Portuguese, provides an example of a prototypically peripheral language with regard to perfect use: the ter perfect is infrequent, semantically archaic in its reference to a present resultative state, and syntactically idiosyncratic in its use of an auxiliary which, by its nature, brings a permansive connotation to the construction, and which may itself represent archaism, if its resemblance to similar Judeo-Spanish constructions can be interpreted as such.

4. Northwestern Spanish varieties

Among the Spanish varieties, Galician will be examined first, as bearing the most formal similarity to Portuguese, with its use of the ter auxiliary. Not unlike Portuguese, the Galician ter perfect is rare, and in Galician, other compound tenses are rare, as well; in both varieties, the synthetic preterite is far more frequent than ← 304 | 305 → the periphrastic perfect.25 Although Harris (1982: 55–56, ftnote 4), and, following him, Squartini / Bertinetto (2000: 410), claim that durative usage is not found in Galician, Paiva Boléo (1936: 14) and Rojo (1974) do find examples of durative usage. In fact, according to Rojo (1974: 122), the perfect in ter always connotes iterativity, whether explicitly, as in (25), or implicitly, as in (26):

(25) que dona Amelia tiña dito moitas vegadas (from A xente da Barreira 1951: 101) ‘which Dona Amelia had said many times’

(26) Esa era outra cousa na que teño cavilado (from Memorias dun neno labrego 1968: 88) ‘This was another thing that I have (repeatedly) pondered

As similarly witnessed in Old Portuguese, haber was frequently used as a perfect auxiliary in many medieval Galician texts, but this usage has virtually disappeared (Rojo 1974: 135); some examples of haber are still to be found in modern written and spoken Galician, but these usages are, according to Rojo (1974: 123), decidedly influenced by Castilian:

(27) Á lus do candil (1968: 61)

Había deixado os estudios.

‘I had left my studies.’

Some instances of Galician ter also appear to have been built on the model of Castilian haber:

(28) A orella no buraco (1965: 66)

Anteriormente tiña sido verde, agora é moura, denegrida, marrón, arrubiscada

‘Previously it had been green, now it is dark, blackened, brown, ruddied’

(examples from Rojo 1974: 128)

The construction tiña sido is based squarely on Castilian había sido ‘had been’, since the expected Galician usage is, rather, the preterite fora (Rojo 1974: 128). Paiva Boléo (1936: 15) also notes that the iterative nuance of the form is fading under Castilian influence.

Galician, then, like Portuguese, shows signs of being extremely conservative in its perfect usage, adhering to the same “peripheral” model: it relies especially on preterites rather than perfects for past situations. When a periphrastic HAVE form is used, the auxiliary tends to retain its archaic meaning, the iterative sense of ter ‘hold, keep’, alongside its perfect meaning. ← 305 | 306 →

Directly east of Galicia, the dialects of Asturia and León share with Galician a low frequency and restricted use of the periphrastic perfect (Paiva Boléo 1936: 16–19; Harris 1982: 53). However, as reported in Harre (1991: 155–59), tener + PP is used more frequently in the Asturian of Oviedo than in either Portuguese or Castilian, and, besides referring to durative or iterative situations which include the present, like Portuguese, it can also refer to experiential situations and situations which exclude the present, unlike Portuguese.

5. Judeo-Spanish

An additional variety which follows the western pattern of perfect formation is the vernacular language of the Jewish community, Judeo-Spanish, or Ladino.26 This variety provides crucial evidence as to the status of the perfects in medieval Spain, since it has retained many archaic features dating back to pre-classical Spanish (Berschin 1976: 142). The present perfect of Judeo-Spanish closely resembles that of Portuguese, Galician, Asturian, and Leonese, not only in its use of tener as the auxiliary, but also in the semantic value of the construction, in consistently denoting current relevance and durative or iterative value (Penny 2000: 189–90). Malinowski (1984: 212) speculates that many Spanish Jews may have lived, before their expulsion, in Galicia and neighboring regions. One might be tempted to suggest, rather, that the use of tener in Judeo-Spanish varieties may indicate more widespread use of this form across medieval Al-Andalus, whether among Arabs, Jews, or Mozarabic Christians, and that this usage came to be adopted when a perfect was called for in Portuguese and the northwestern varieties. In all of these varieties, it should be noted, the preterite shows much higher frequency than the periphrastic perfect.

6. The emergence of Castilian dominance

During the early years of Moorish predominance in the south, León in the west and Aragón in the east were the most important Christian kingdoms, but under the rule of Fernando I (1035–1065) and Alfonso VI (1065–1109), Castile came to be the most important Spanish Christian kingdom. Both of these rulers imported monastic traditions and clerical reformations from the north, but it ← 306 | 307 → was Alfonso who in 1080 made the effort referred to above to establish greater uniformity with other Christian lands of Europe by introducing, at the behest of Pope Gregory VII, the Roman liturgy to replace the Mozarabic rite. Alfonso also appointed, as his first metropolitan of Toledo, not a Mozarabic but a Frankish Cluniac clergyman (Menéndez Pidal 2005: 409; 417). Even more significantly, of all monarchs in Europe, Fernando and Alfonso were the most prolific financial supporters of Cluny Abbey itself, funding the expansion of the monastery with annual donations amounting to some 100,000 deniers obtained as booty from Muslim conquests, to the extent that Cluny became the largest church in the Christian world (Gerhards 1992: 49–50). Such an alignment with this renowned symbol of Frankish prestige and power would, without doubt, foster cultural and linguistic realignments, as well, and the perfects, like other linguistic artifacts, attest to this flow of influence.

In early times, Castilian shared a number of features with western varieties such as Portuguese, Galician, Asturian, and Leonese. For example, 13th-century Castilian perfects tended towards iterative or durative usage much more frequently than the synthetic preterites did (Harre 1991: 115). But as time went on, Castilian oriented itself more and more to the eastern varieties, especially from the 15th to 17th centuries, clearly as a result of the political union of Castile and Aragón (Meier 1970: 610). With regard to the impact of Castilian political power on the development of the periphrastic perfect, Castilian provided a model which a number of other varieties copied, with its unified use of haber as perfect auxiliary, at the expense of ser, its retention of participle agreement, and in other features of perfect usage (Westmoreland 1988: 383).

6.1 Castilian BE

As we observed in Portuguese, the auxiliary ser ‘be’ was used more productively in earlier times, but has disappeared from Modern Spanish usage. Ser was used in Old Spanish for unaccusatives such as ir ‘go’, venir ‘come’, llegar ‘arrive’, finir ‘finish’, morir ‘die’ (Benzing 1931). In the 12th century, however, ser was still used for unaccusatives (29), but not for true reflexives (30):

(29)Cantar de Mio Cid(late 12th -early 13th c.)
   somos vengados ‘we have (lit. are) taken revenge
(30)*me so(y) labado ‘I have washed myself’
(Tuttle 1986: 264)

From the 13th century onward, HAVE steadily increased at the expense of the BE auxiliary, to the point of virtually ousting BE by the 16th century (Pountain 1985: 337; Harre 1991: 116). Figure 3 illustrates this lengthy decline in the use of ser ← 307 | 308 → with intransitive verbs, reaching the level of 50 % in El Corbacho (1438) and 20 % in La Celestina (1499).

Figure 3: The use of ser and haber with intransitive verbs (from Rodríguez Molina 2010: 1030 after Company Company 1983)27


Rodríguez Molina (2010: 1036–37) points out that the meaning of ser + PP is not equivalent to that of the haber + PP construction, and that the ser construction is therefore better characterized as a resultative rather than an anterior. Following Rodríguez Molina (2010), Rosemeyer (2014: 71; 2015: 312) concludes that it was the grammaticalization of haber + PP that was ultimately responsible for the increased productivity of the haber construction and its gradual expansion of type frequency at the expense of the ser construction.28 Rodríguez Molina (2010) also ← 308 | 309 → points to the introduction of periphrasis itself as favoring the expansion of the HAVE category to the detriment of the BE category.

What is claimed here, then, is that the Spanish BE + PP construction, inherited from Latin, did not undergo the reinvigoration as a perfect category that the varieties within the Charlemagne Sprachbund did, but, rather, underwent a gradual decrease in frequency to the point of complete loss.

6.2 Castilian HAVE

In the 13th century, the synthetic perfect clearly referred to past action, as in (31), while the much rarer periphrastic perfect referred to a state or present result of a past action, as in (32) (Alarcos Llorach 1973: 40–41):

(31) Libro de Alexandre line 1265

Dixieste grant basemia

‘you spoke a large curse’

(32) Libro de Alexandre line 350

Mucho me as bien fecho

‘you have done much for me’

At this early time, the periphrastic perfect could also include reference to the iterative or durative nature of the action which produced this result (Harre 1991: 115):

(33) Libro de Alexandre line 351

Tanto avemos fecho que los dios son yrados

‘we have done so much that the gods are enraged’ (example from Alarcos Llorach 1973: 41)

(34) Primera crónica general: G 14, 11–13L (2nd half of the 13th century)

E fizieron por toda la tierra cantares de llantos dEspanna que dizien que Dios la auie ayrada

‘And throughout the land they performed laments for Spain, for they said that God was angry with her’ (example from Harre 1991: 115)

Harre (1991: 115) notes, however, that this durativity was not found in any of the later texts she examined.

As illustrated above in Table 1 and Figure 2, the 15th century saw a rather sharp rise in the use of Castilian haber at the expense of tener. According to Harre (1991: 121), this increase is to be found especially from the time of La Celestina onwards, that is, shortly after 1500. It will be worth our while to take a closer look at the evidence provided by La Celestina, since its stratified nature provides clear documentation of several developments in the Castilian verb system at the beginning of the 16th century: while Act 1 retains many archaic characteristics, Acts 2–16, dating to a later period and written by a different author, show many ← 309 | 310 → innovative features. Using the analysis of Criado de Val (1955), we can note the following trends:

1. As noted above, the synthetic preterite remains more frequent than the perfect throughout the work,29 but the use of the periphrastic perfect is increasing: the perfect is not well attested in Act 1, but grows in use in the later acts (Criado de Val 1955: 95; 101; 212).30

2. The primary marker of possession shifts from haber to tener. In Act 1, haber can refer to the possession of material objects, but in the later acts, it is used almost exclusively for abstract or hypothetical possession. As tener becomes the general verb of possession in the later acts, it also tends to show an iterative sense, as in many modern varieties. As Criado de Val notes (1955: 60), haber persists as the preferred marker of possession in cultured writing of the 14th and 15th centuries, but tener emerges in the dialog of the play as the preferred vernacular form. The growth of tener is also documented in other works of the time (Criado de Val 1955: 50). And as tener replaced haber as the independent possessive form, haber took on a stronger auxiliary function (Alarcos Llorach 1947: 131; 1973: 40).

3. Concomitant with the haber / tener shift is the replacement of ser by estar in many contexts over the course of the century, establishing a parallel set of aspectual distinctions across the auxiliary systems.

4. With regard more precisely to the use of these verbs as perfect auxiliaries, haber begins to replace ser + PP for unaccusative verbs, as found in Modern Castilian. In Act 1, no examples of haber appear with unaccusative verbs; in the later acts, on the other hand, many examples occur, alongside ser (Criado de Val 1955: 63–65):


Act 1 Celestina:Que, avnque á vn fin soy llamada, áotro so venida.

‘For, although I have been called here for a purpose, I have [lit. am] come for another’


Act 4 Melibea:¿Por quien has venido ábuscar la muerte para ti?

‘For whom have you gone looking for your death?’ ← 310 | 311 →

5. Likewise, a more productive use of tener as an auxiliary emerges in the later acts:

Act 4: Melibea: ¿Piensas que no tengo sentidas tus pisadas y entendido tu dañado mensaje?
 ‘Do you think that I have not heard your footsteps and understood your damned message?
 (i.e., ‘Do you think that I cannot see what you are trying to do?)
Act 10: Celestina:Yo lo tengo pensado
 ‘I have thought about it’
Act 10: Melibea:¿E no te tengo dicho que no me alabes esse hombre?
 Have I not told you not to praise that man?

The use of tengo dicho is especially frequent in the later acts. On the other hand, there is only one example of tener used as an auxiliary in Act 1. Thus, at the beginning of the 16th century, haber and especially ser grew more restricted across a range of uses, and came to be replaced by tener and estar in many contexts. In spite of this increase, haber still predominates over other auxiliaries in periphrastic constructions (Criado de Val 1955: 183).

The 16th century sees the transformation of perfect auxiliation in Castilian: haber comes to be consolidated as the perfect auxiliary not only through the above-mentioned decline in the use of ser in this role, but also because of the growing tendency for tener to mark possession at the expense of haber.

In sum, Castilian provides evidence of innovation and grammaticalization within its periphrastic perfect, but the perfect remains generally conservative in meaning and “peripheral” in form, having given up the BE auxiliary in favor of HAVE, and remaining considerably less frequent than the synthetic past.

Before leaving our examination of Castilian, brief mention should also be made of the crucial evidence provided by the Spanish varieties of North and South America. Significantly, many of these New World varieties retain distinctly archaic patterns, showing a decided preference for the synthetic preterite over the periphrastic perfect, and a tendency for perfects to connote durativity or iterativity, not unlike the patterns to be witnessed in present-day Peninsular Portuguese and in Peninsular Spanish of the early Renaissance, illustrated above.31 With regard to the predominance of synthetic preterites, Rodríguez Louro (2010, Table 2) shows ← 311 | 312 → that modern speakers of Argentinian and Mexican Spanish use the preterite 90 % and 85 % of the time, respectively, as opposed to present-day speakers of Peninsular Spanish, who use it 46 % of the time. As regards the permansive value of the perfects, this tendency has been widely noted in Mexico, Colombia, Argentina32, Puerto Rico, and the Canaries33 (see Westmoreland 1988 and Squartini / Bertinetto 2000: 412–13 for extensive references). Such evidence allows us not only to gain access to the nature of the Spanish perfects of an earlier time, but also to recognize the fact that change does not proceed in one direction alone or at a uniform pace: in their retention of older patterns, the Spanish varieties of the Americas demonstrate even more peripherality than those of the Iberian Peninsula; in their adoption of patterns from other varieties, such as evidential semantic value of perfects from Quechua (Howe 2013) or the newer hodiernal interpretation from Castilian, on the other hand, they illustrate the essential role that contact plays in determining which path innovations will follow.

7. Aragonese

Moving eastward, we recognize several features in Aragonese which, as noted above (1.1), suggest that this is the point of origination for the HAVE + PP construction on the Iberian peninsula, and that this innovation was introduced through Carolingian influence. It is precisely here, as well as in Navarra and in the Basque territory, where the highest frequency of HAVE + PP appears prior to 1400, and where ← 312 | 313 → the most advanced grammaticalization of this structure has occurred, appearing, for example, with intransitive verbs more frequently than elsewhere, with non-canonical objects, and with modals (Rodríguez Molina 2010: 1217–19).

Alongside these indications of the early establishment of the HAVE construction can also be placed the more frequent use of the BE auxiliary in Aragonese perfects, in contrast to its very limited use in Castilian. The BE / HAVE distinction, similar to the French distribution, persists in some varieties of Aragonese even to the present day.34 Also like French is the fact that the verb haber could serve both as the auxiliary and as the verb of possession (Umphrey 1913: 39; 45). Moreover, in some dialects of modern Aragonese, just as in Old Aragonese, both haber (37) and ser (38) persist as main verbs, in opposition to Castilian’s adoption of tener and estar to mark general possession and existence. All of these tendencies recall the Carolingian distribution.35

(37)Old Aragonese possessive ‘have’el rio que ha nombre Ana
 ‘the river that has the name Ana’
(38)Old Aragonese auxiliary BEde todos tiempos era seydo vencedor
 ‘of all times, he had been the winner’
  (Kuhn 1935: 154–5)

Aragonese is likewise much more conservative than Castilian in maintaining agreement between the object and the participle. The letters between King Pedro IV of Aragón (reigned 1336–1387) and King Pedro I of Castile (reigned from 1350–1366 and 1367 to 1369) illustrate this difference well:

(39)Letter from the King Pedro IV of Aragón:
   your.F.SG letter.F.SG have.1PLreceive.PP.F.SG
   ‘we have received your letter’ ← 313 | 314 →
(40)Letter from the King Pedro I of Castile:
       losde nuestros regnosavemosrecibido
       the.M.PL of our.M.PL reign.M.PLhave.1PLreceive.PP.M.SG
       ‘we have received those of our reigns’
       (examples from Umphrey 1913: 38)

There are, in fact, no examples of participial agreement in the letters of the King of Castile examined by Umphrey (1913: 38), but, in one letter alone of the King of Aragón, there are nine examples of agreement, and only one instance of non-agreement.

Latin documents written in Aragón also demonstrate a shifting orientation towards Carolingian tradition through the adoption of Caroline script in place of Visigothic. In the first half of the 12th century, for example, the chancery used Visigothic script, while the ecclesiastical centers used Caroline. Remarkably, the transition to the Caroline script is captured in statu nascendi in the writing of a particular scribe who wrote in the Visigothic style until May, 1147, when he switched to the Caroline script (Balaguer 1954). This innovation is probably tied to the unification of Aragón and Catalonia in 1137 into the Corona de Aragón (Wright 1982: 235). By 1200, the Visigothic script had been given up all over the Iberian Peninsula, persisting only in Toledo, a gathering point for many Christians exiled from Muslim territories who had not been exposed to the cultural and liturgical influences from the north.

8. Catalan

Turning our attention to Catalan, we recognize a strong connection with the core of western Europe and with the Carolingian Empire: part of the territory was briefly occupied by the Muslims, but reclaimed by Carolingian forces in 778, and was incorporated into the Carolingian holdings in 801, when Charlemagne’s son, Louis the Pious, captured Barcelona. Catalonia thus became part of the Spanish March, a military buffer between Al-Andalus and the Carolingian Empire. Its connection with the Empire during the time of Charlemagne’s rule is duly reflected in its productive early use of both BE and HAVE auxiliaries, following patterns similar to those found in French and northern Italian (Lapesa 1968; Squartini / Bertinetto 2000).

While there are no chronicles or other narrative historical sources written in Catalan before the 12th century to document the history of the area and its connection with the Carolingian Empire, official documents written in the vernacular appear as early as the 9th century, and a profusion of administrative documents in Latin survives, the richest collection of such documents in 9th century medieval ← 314 | 315 → Europe outside of northern Italy (Collins 1990: 172). The abundance of documents is a reflex of Catalan’s complex history: unlike the rest of the Carolingian realm, Catalonia continued to adhere to Visigothic Law, the Lex Gothorum, which, ultimately, represented a continuation of Late Roman practice. Vestiges of this ancient tradition can still be found in the Catalan commitment to written documentation of all legal proceedings, and in the persistent need to seek the king’s confirmation for all land entitlements. These two tendencies, taken together, explain why so many Catalan abbots, bishops, and landowners traveled to the Carolingian court over the 9th and 10th centuries, and why the ties with the Carolingian realm continued to be so strong (Collins 1990: 181–2; 184–5).

When we examine these Latin documents, many of which are collected in Abadal i de Vinyals (1926–1950; 1952), we realize that, while few provide evidence for the vernacular use of the periphrastic perfect, since the synthetic preterite appears pervasively, the best evidence for the perfect comes from the late 8th and early 9th centuries, that is, during the reign of Charlemagne himself, as the following two examples from the same precept illustrate (41). The letter concerns complaints from settlers who were given “aprisions”, i.e., land grants to settle the “squalid” unsettled lands of the Spanish Marches. The settlers, called “Hispani,” were often refugees from Spain, sometimes with Arabic names, and had often served in the military or as mercenaries in the borderlands.36 Their settlement in the deserted territories of Catalonia helped provide the buffer which the Carolingians desired as defense against incursions from the Ummayad (Collins 1990: 182–86):

(41)Charlemagne’s Præceptum pro Hispanis, April 2, 812, concerning aprisions (Abadal i de Vinyals 1952: 314)
 laboratas[ab]illiseisabstractas habeatis
 work.PP.NOM.F.PLfromthose theseseparate.PP.NOM.F.PL have.2PL.SBJ
 et     quicquid contra   justiciam eisvos aut juniores vestri
 and whatever against justicetheseyou or younger your
 do.PP.ACC.M.SG have.2PL.PRES
 ‘They also say that some farms, which they themselves worked, you have worked and separated from these and whatever you or your heirs have done against justice’ ← 315 | 316 →

A further example is found in the precept of Charlemagne’s son, Louis, when he first inherited his father’s throne:

(42)Louis’s Præceptum, January 1, 815 (Abadal i de Vinyals 1952: 321)
       etquicquidille occupatumhabebataud aprisione fecerat
             and whatever this occupy.PP.NOM.M.SG have.3S.IMPRF. or aprision
             ‘and whatever he had occupied and had made as a land grant’

While habere is often to be found in the documents collected in this volume as a possessive verb or in frozen forms (licentium etiam habeant ‘they also have.3PL.SBJV license’), I did not find other examples of its use as an auxiliary here. BE auxiliaries do exist among these documents, retaining their passive sense (e.g., 878 memorata sunt ‘was mentioned’; dedicata est ‘was dedicated’ Abadal i de Vinyals 1952: 70). The synthetic form was greatly preferred to the periphrastic, probably as representing the more authentic classical Latin choice.

A further overt sign of the movement of Carolingian influence from east to west can be seen in the adoption of the Roman liturgy and Caroline miniscule script first in the eastern part of Catalonia, early in the ninth century, and only later in the western part, in the tenth century. We can also view the presence of bilingual versions of sermons, where the vernacular Catalan appears alongside the Latin, as a sign of adoption of Carolingian traditions, since these resemble such materials in other parts of the Carolingian territory (Wright 1982: 150). The presence of such bilingual texts could also give us some clue as to how Latin structural patterns could easily influence those of the vernacular.

The marginal status of Catalan in the Empire is likewise represented by its shifting auxiliary patterns: as we have noted, Catalan took part in Carolingian trends early in its history, using BE as an auxiliary even in true reflexives as in (43):

(43) Fr. Antoni Canals, (14th c.)

après que Cató se fou gitat sobre la sua espasa

‘after C. had (lit. was) thrown himself on his sword’

The use of BE auxiliation with verbs of motion persists in some varieties of Catalan to this day, as shown in (44):

(44) Catalan (Badía Margarit 1951: 326)

                   be.1SG come.PP
                   ‘I have [lit. be] come← 316 | 317 →

However, since the 15th century, many varieties have gravitated toward the Castilian pattern of HAVE-predominance.37 Catalan’s reorientation towards other varieties of the Iberian Peninsula can also be seen elsewhere in its verb system: in the semantic extension of tenir from ‘hold’ to its use as the unmarked verb of possession, with the concurrent relegation of haver to auxiliary-only status, and, likewise, in the later parallel establishment of estar as copula, impinging upon ser (Steinkrüger 1995).38 Nevertheless, the fact that, by the beginning of the 16th century, many varieties of Catalan had taken part in the semantic shift of perfect to preterite similar to that which occurred in French and northern Italy (Eberenz 1977; Kempas 2014) implies a continued relationship with languages outside the Iberian Peninsula.39

9. Trans-Pyrenean influence

A final piece of Iberian evidence pointing to the role that the Carolingian Empire played in the reinforcement of HAVE and BE auxiliation may, somewhat surprisingly, be found in the earliest surviving piece of written Spanish literature, the Auto de los Reyes Magos, dating to the end of the twelfth century. After the capture of Toledo in 1085 by Alfonso VI, the munificent benefactor of Cluny mentioned above, French-based clerics of Cluny were brought in for Latin instruction and the establishment of the Roman rite. It is possible that the author of this work ← 317 | 318 → was one of these clerics—a priest from Gascony (Lapesa 1954; 1967: 143), Catalonia (Kerkhof 1979), or La Rioja (Hilty 2004). Because the work is composed in Castilian, however, a more likely explanation is that the author was a Castilian writer who had been influenced by the diversity of the cultural influx into Toledo (Sánchez-Prieto Borja 2003: 209; Rodríguez Molina 2010: 738).

What we discover from examining the fragments of this early dramatic work, apparently composed to be performed in church for Epiphany, is a wealth of periphrastic perfects already in place. Three appear in the opening lines, and continue like a refrain throughout the play:



CasparDios criador; qual marauilla‘God Creator, what a wonder
no se cual es achesta strela!I do not know what this star is!
Agora primas la he ueida,Now is the first time I have seen it,
poco tiempo a que es nacida.a little time since it was born
Nacido es el Criador.The Creator is born.
Baltasar[….]Certas nacido es […]Surely he is born.
En todo, en todo es nacido?All in all, is he born?
non se si algo e ueido;I don’t know if I have seen anything;
(6 more times es nacido/a) 
HerodesI cumo lo sabedes?And how do you know it?
ia prouado lo auedes?How have you proven it?
CasparRei, vertad te dizremos,King, truly we tell you
que prouado lo auemos.That we have proven it.
MelchiorEsto es grand ma[ra]uila.This is a great wonder.
un strela es nacida.A star is born.
BaltasarSennal face que es nacidoIt makes a sign that he is born
i in carne humana uenido.and in human flesh is come.

Not only is the high frequency of periphrastic perfect use itself noteworthy, but so too is the prominence of participial agreement, used for literary effect in contrasting the birth of the star (feminine) to the birth of Christ (masculine) throughout. This high frequency of perfect use also points to influence from north of the Pyrenees, and positions this earliest Spanish vernacular work as partaking of Carolingian linguistic traditions. This assumption, if correct, leads us to another somewhat paradoxical observation: Toledo, in its role as cultural crossroads, provides evidence not only of conservative tendencies, as seen in the retention of Visigothic scribal traditions noted above, but also of innovative linguistic tendencies introduced by émigrés from the north, as noted here. ← 318 | 319 →

10. Conclusions

The developments of the periphrastic perfects across the Iberian Peninsula are summarized in Table 2, with shading representing connections and influence across varieties:

Table 2: Summary of major developments in perfects of Iberian Peninsula


← 319 | 320 →

Three facts are especially noteworthy:

In Castilian, both early ties to western varieties and later connections to eastern varieties are reflected in the distribution of the perfects

The influential status of Castile is also evident in the fact that Aragonese and Catalan gravitate toward its style of perfect formation, as do Galician and Leonese, to some extent

Portuguese and Castilian, similar in their perfect profiles in medieval times, undergo a definitive split in the 15th–16th centuries: Castilian opts for haber as its only perfect auxiliary, while Portuguese establishes the more vernacular ter.

Pountain (1985: 346) summarizes the latter development as follows:

Portuguese, which shares the early pattern of change with Spanish, seems to have taken the process embryonically present in Spanish—the development of tener-ter as an auxiliary—to its extreme.

Wheatley (1995: 143) expands on this observation, regarding Portuguese as maintaining the trajectory that the two languages had shared, but Spanish as halting the progression that had been set in motion:

the same development may have been occurring in both languages [Portuguese and Spanish], but was somehow impeded in Spanish during this period, never reaching intransitive and reflexive verbs.

Some modification of this claim is apparently in order: since both Portuguese and Spanish showed variability in auxiliary use in medieval times, with haver / haber predominating over ter / tener in both varieties, we should not characterize Portuguese as simply remaining on an ancient trajectory, and Spanish stopping this progression. Rather, we should view both Portuguese and Spanish as moving in different directions from their similar variable starting points. Significantly, the marking that was chosen in Portuguese reflects vernacular tendencies that appear to have been influenced by contact with Arabic, while the marking that was chosen in Castilian coincides, at least to some extent, with the patterns to be found in the eastern Iberian varieties and those north of the Pyrenees.

One final observation bears mentioning in connection with the role that Arabic may have played in the development of the Romance varieties of the Iberian Peninsula. Wasserstein (1991: 13), in his discussion of the effect of the widespread use of Arabic in both speaking and writing, notes that

during the eleventh century, the efflorescence of Arabic culture in Spain was accompanied by the first real advances of the Christian reconquest. One effect of the Christian successes may have been a hardening of the linguistic differences, as one marker of the group boundaries that came to be more strongly etched in Iberian life at precisely this time. ← 320 | 321 →

Just as the hardening of the boundaries may have occurred in the Arabic-speaking community, so might this reinforcement of boundaries have been experienced in the language of the Reconquistadores. As we have seen, some of the changes which occurred in Castilian took place precisely at the time of the reconquest. While Castilian does not copy the BE / HAVE alternation found in eastern varieties like Aragón and Catalan, which partake more fully of Carolingian trends, it does reinforce its perfects formed with haber at this time, in contrast to the diminution of the category in Portuguese, Galician, and in other northwestern varieties. It is not impossible that a reaction to southern and western traditions of perfect-formation was taking place, as a means of demarcating and reinforcing a significant cultural boundary.

What we conclude, then, with Rodríguez Molina (2010), is that the development of the periphrastic perfect on the Iberian Peninsula is strongly dependent on historical and cultural trends and on geographical locale. What we add to Rodríguez Molina’s findings, however, is evidence that, in addition to Frankish influence upon the structural patterns of the eastern varieties, it may also be possible to identify the effects of Arabic influence on the structures of the west.


Abadal i de Vinyals, Ramon d’ (1926–1950 / 1952): Catalunya carolingia. Vol. 2: Els diplomes carolingis a Catalunya (Part I: 1926–1950; Part II: 1952), Barcelona: Institució Patxot.

Alarcos Llorach, Emilio (1947): “Perfecto simple y compuesto en español”, Revista de Filología Española 31, 108–139.

Alarcos Llorach, Emilio (1973): Estudios de gramática funcional del español, Madrid: Gredos.

Algeo, James (1976): “The Portuguese present perfect”, Luso-Brazilian Review 13, 194–208.

Amman, Andreas (2002): “Arabic verbal inflection: An essay in de-exoticizing”, Sprachtypologie und Universalienforschung 55, 311–339.

Badía Margarit, Antonio (1951): Gramática histórica catalana, Barcelona: Noguer.

Balaguer, Federico (1954): “Una nota sobre la introduccion de la letra carolina en la cancillería aragonesa”, Jeronimo Zurita. Cuadernos de Historia 3, 155–161.

Beale-Rosano-Rivaya, Yasmine Consuelo (2006): Mozarabic : culture, contact, language and diglossia, dissertation, Los Angeles: UCLA.

Benzing, Joseph (1931): “Zur Geschichte von ser als Hilfszeitwort bei den intransitiven Verben im Spanischen”, Zeitschrift für Romanische Philologie 51, 385–460. ← 321 | 322 →

Berschin, Helmut (1976): “Präteritum und Perfektgebrauch im heutigen Spanisch”, Tübingen: Niemeyer (Beihefte der Zeitschrift für Romanische Philologie, 157).

Brinkmann, Hennig (1931): Sprachwandel und Sprachbewegungen in althochdeutscher Zeit, Jena: Biedermann.

Bustamante, Isabel (1991): “El presente perfecto o pretérito perfecto compuesto en el español quiteño”, Lexis 15, 195–231.

Carballo Calero, Ricardo (21970): Gramática elemental del gallego común, Vigo: Galaxia.

Collins, Roger (21990): “Charles the Bald and Wilfred the Hairy”, in Margaret Gibson and Janet Nelson (eds.), Charles the Bald: Court and Kingdom, Farnham: Ashgate, 169–88. [reprinted in Collins, Roger (1992): Law, Culture and Regionalism in Early Medieval Spain, Hampshire: Variorum]

Company, Concepción (1983): “Sintaxis y valores de los tiempos compuestos en el español Medieval”, Nueva Revista de Filología Hispánica 32, 235–257.

Corriente, Federico (1991): “The behaviour of Romance and Andalusian utterances in the kharjas under the constraints of ‘arūd”, in Jones / Hitchcock (eds.), 60–67.

Corriente, Federico (1997): Poesía dialectal árabe y romance en Alandalús, Madrid: Gredos.

Criado de Val, Manuel (1955): Índice verbal de la Celestina, Madrid: CSIC.

Cuvalay-Haak, Martine (1997): The verb in Literary and Colloquial Arabic, Berlin / New York: De Gruyter.

Dahl, Östen / Velupillai, Viveka (2013): “Perfects”, in Matthew Dryer and Martin Haspelmath (eds.), The World Atlas of Language Structures Online, Leipzig: Max Planck Institute for Evolutionary Anthropology [online], <>.

Detges, Ulrich (2004): “How cognitive is grammaticalization? The history of the Catalan perfet perifràstic”, in Olga Fischer and Muriel Norde (eds.), Up and down the cline: The nature of grammaticalization, Amsterdam / Philadelphia: John Benjamins, 211–227.

Drinka, Bridget (2013): “Sources of auxiliation in the perfects of Europe”, in Freek van de Velde et al. (eds.), Multiple source constructions in language change (special issue of Studies in Language 37: 3), 599–644.

Drinka, Bridget (forthcoming): Language Contact in Europe: The Periphrastic Perfect through history, Cambridge: Cambridge University Press.

Eberenz, Rolf (1977): “Zur Entwicklung der Opposition perfecto simple vs. perfecto compuesto im Spanischen und Katalanischen der Renaissance”, Zeitschrift für romanische Philologie 93: 518–527. ← 322 | 323 →

Escobar, Anna María (2000): Contacto social y lingüistico: el español en contacto con el quechua en el Perú, San Miguel: Pontificia Universidad Católica del Perú.

Fasi Fehri, Abdelkader (2003): “Arabic Perfect and temporal adverbs”, in Artemis Alexiadou et al. (eds.), Perfect explorations, Berlin / New York: De Gruyter, 69–99.

Flobert, Pierre (1975): Les verbes déponents latins des origines à Charlemagne, Paris: Les Belles Lettres.

Gerhards, Agnès (1992): L’Abbaye de Cluny, Paris: Éditions Complexe.

Gil, Juan (1971): “Apuntes sobre la morfologia de Albaro de Córdoba”, Habis 2, 199–206.

Gil, Juan (1973): Corpus Scriptorum Muzarabicorum, vol. I, Madrid: Instituto Antonio de Nebrija.

Gillmann, Melitta (2015): “Auxiliary selection in closely related languages: the case of German and Dutch”, in Kailuweit / Rosemeyer (eds.), 333–358.

Glick, Thomas (1979): Islamic and Christian Spain in the Early Middle Ages, Princeton: Princeton University Press.

Harre, Catherine (1991): Tener + Past Participle. A case study in language description, London / New York: Routledge.

Harris, Martin (1982): “The ‘past simple’ and the ‘present perfect’ in Romance”, in Nigel Vincent and Martin Harris (eds.), Studies in the Romance verb, London / Canberra: Croom Helm, 42–70.

Haxen, Ulf (1991): “Kharjas in Hebrew Muwassahs”, in Jones / Hitchcock (eds.), 37–48.

Herrera Santana, Juana / Medina López, Javier (1991): “Perfecto simple / perfecto compuesto: análisis sociolingüístico”, Revista de Filología de la Universidad de La Laguna 10, 227–239.

Hourani, George (1972): “The Medieval translations from Arabic to Latin made in Spain”, The Muslim World 62, 97–114.

Howe, Chad (2013): The Spanish perfects: Pathways of emergent meaning, Houndsmills: Palgrave Macmillan.

Huber, Joseph (1933): Altportugiesisches Elementarbuch, Heidelberg: Winter.

Jones, Alan (1991): “Omnia vincit amor”, in Jones / Hitchcock (eds.), 88–103.

Jones, Alan / Hitchcock, Richard (eds.) (1991): Studies on the muwaššah̟ and the kharja, Reading: Ithaca Press.

Kailuweit, Rolf (2015): “BE or HAVE in Contemporary Standard French —residua of semantic motivation”, in Kailuweit / Rosemeyer (eds.), 249–275. ← 323 | 324 →

Kailuweit, Rolf / Rosemeyer, Malte (eds.), Auxiliary selection revisited: gradience and gradualness, Berlín / Nueva York: De Gruyter.

Kempas, Ilpo (2014): “Aportes empíricos a los estudios sobre el pretérito perfecto compuesto ‘aoristizado’”, in Susana Azpiazu (ed.), Formas simples y compuestas de pasado en el verbo español, Lugo: Axac, 81–102.

Kinberg, Naphtali (1992): “Semi-imperfectives and imperfectives: a case study of aspect and tense in Arabic participial clauses”, Lingua 86, 301–330.

Kuhn, Alwin (1935): “Der hocharagonesische Dialekt”, Revue de Linguistique Romane 11, 1–312.

Lapesa, Rafael (1954): “Sobre el Auto de los Reyes Magos: sus rimas anómalas y el posible origen de su autor”, in Homenaje a Fritz Kruger, vol. II, Mendoza: Universidad Nacional de Cuyo, 591–599.

Lapesa, Rafael (71968): Historia de la lengua española, Madrid: Escelicer.

McCoard, Robert (1978): The English perfect: tense-choice and pragmatic inferences, Amsterdam / New York / Oxford: North-Holland.

McKitterick, Rosamond (2008): Charlemagne: the formation of a European identity, Cambridge: Cambridge University Press.

Malinowski, Arlene. 1984 [1989]. “Distribution and function of the auxiliaries tener and aver in Judeo-Spanish”, Orbis 33, 211–221.

Matoso Câmara, Joaquim (1972): The Portuguese language, Chicago: University of Chicago Press.

Meier, Harri (1970): “Sintaxis verbal española, peninsular e hispanoamericana”, in Carlos Magis (ed.), Actas del Tercer Congreso Internacional de Hispanistas 1968, Mexico: El Colegio de México, 601–610.

Mendeloff, Henry (1968): “The Catalan periphrastic perfect reconsidered”, Romanistisches Jahrbuch 19, 319–326.

Menéndez Pidal, Ramón (2005): Historia de la lengua española, edited by Diego Catalán, Madrid: Fundación Ramón Menéndez Pidal / Real Academia Española.

Menjot, Denis (1995): “Brassages linguistiques et langue écrite: le castillan dialectal de Murcie au Bas Moyen Âge”, Le Moyen Âge 101, 447–459.

Norberg, Dag (1943): Syntaktische Forschungen auf dem Gebiete des Spätlateins und des Frühen Mittellateins, Uppsala: Lundequistiska Bokhandeln.

Penny, Ralph (2000): Variation and change in Spanish, Cambridge: Cambridge University Press.

Paiva Boléo, Manuel (1936): O perfeito e o pretérito em português em confronto com as outras linguas românicas, Coimbra: Biblioteca da Universidade.

Paterson, Linda (1993): The world of the troubadours: Medieval Occitan society c. 1100 – c. 1300, Cambridge: Cambridge University Press. ← 324 | 325 →

Pountain, Christopher (1985): “Copulas, verbs of possession and auxiliaries in Old Spanish: the evidence for structurally interdependent changes”, Bulletin of Hispanic Studies 62, 337–55.

Rodríguez Louro, Celeste (2010): “Past time reference and the Present Perfect in Argentinian Spanish”, in Yvonne Treis and Rik de Busser (eds.), Selected papers from the 2009 Conference of the Australian Linguistic Society, Torino: Einaudi, 1–25.

Rodríguez Molina, Javier (2010): La gramaticalización de los tiempos compuestos en español antiguo: cinco cambios diacrónicos, dissertation, Madrid: Universidad Autonoma de Madrid.

Rojo, Guillermo (1974): Perífrasis verbales en el gallego actual, Santiago de Compostela: Universidad de Santiago de Compostela (Anejos de Verba, 2).

Rosemeyer, Malte (2014): Auxiliary selection in Spanish: Gradience, gradualness, and conservation, Amsterdam / Philadelphia: John Benjamins.

Rosemeyer, Malte (2015): “Entrenchment and discourse traditions in Spanish auxiliary selection”, in Kailuweit / Rosemeyer (eds.), 301–331.

Sánchez-Prieto Borja, Pedro (2003): “¿Rimas anómalas en el Auto de los Reyes Magos?”, Revista de Literatura Medieval 16, 149–219.

Scholz, Bernhard (1970): Carolingian Chronicles, Ann Arbor: University of Michigan Press.

Schumacher de Peña, Gertrude (1980): “El pasado en español andino de Puno / Perú”, en Hans Dieter Bork et al. (eds.), Romanica Europea et Americana. Festschrift für Harri Meier, Bonn: Bouvier, 553–558.

Serrano, María José (1995): “Sobre el uso del pretérito perfecto y pretérito indefinido en el español de Canarias: pragmática y variación”, Boletín de Filología de la Universidad de Chile 35, 527–60.

Smith, John Charles (1989): “Actualization reanalyzed: Evidence from the Romance compound past tenses”, in Thomas Walsh (ed.), Synchronic and diachronic approaches to language variation and change, Washington: Georgetown University Press, 310–325.

Squartini, Mario / Bertinetto, Pier Marco (2000): “The simple and compound past in Romance languages”, in Östen Dahl (ed.), Tense and aspect in the languages of Europe, Berlin / New York: De Gruyter, 403–439.

Steinkrüger, Patrick (1995): ”Grammatikalisierungen von Auxiliaren und Copulae im Katalanischen der Decadència”, Zeitschrift für Katalanistik 8: 35–62.

Sten, Holger (1973): L’emploi des temps en portugais moderne, Copenhagen: Munksgaard.

Suter, Alfred (1984): Das portugiesische Pretérito Perfeito Composto, Bern: Francke. ← 325 | 326 →

Thieroff, Rolf (2000): “On the areal distribution of tense-aspect categories in Europe”, in Östen Dahl (ed.), Tense and aspect in the languages of Europe, Berlin / New York: De Gruyter, 265–305.

Thibault, André (2000): Perfecto simple y perfecto compuesto en español preclásico: estudio de los perfectos de indicativo en La Celestina, el teatro de Encina y el Dialogo de la lengua”, Tübingen: Niemeyer.

Thielmann, Philip (1885): “Habere mit dem Part. Perf. Pass.”, Archiv für Lateinische Lexikographie und Grammatik 2, 372–423 / 509–549.

Tuten, Donald (2003): Koineization in Medieval Spanish, Berlin / New York: De Gruyter.

Tuttle, Edward (1986): “The spread of ESSE as universal auxiliary in Central Italo-Romance”, Medioevo romanzo 11, 239–287.

Umphrey, George (1913): ”The Aragonese dialect”, Extrait de la Revue Hispanique 24, 5–45.

van der Auwera, Johan (1998): “Conclusion”, in Johan van der Auwera (ed.), Adverbial constructions in the languages of Europe, Berlin /New York: De Gruyter, 813–836.

Vanhove, Martine (2001): “Contacts de langues et complexification des systèmes: le cas du maltais”, Faits de Langues 18, 65–74.

Wasserstein, David (1991): “The language situation in al-Andalus”, in Jones / Hitchcock (eds.), 1–15.

Westmoreland, Maurice (1988): “The distribution and the use of the Present Perfect and Past Perfect forms in American Spanish”, Hispania 71: 379–384.

Wheatley, Kathleen (1995): The grammaticalization of the perfect periphrasis in Medieval Spanish and Modern Romance, dissertation, Ann Arbor: University of Michigan.

Wright, Roger (1982): Late Latin and early Romance in Spain and Carolingian France, Liverpool: Francis Cairns.

Wright, Roger (1994): Early Ibero-Romance: Twenty-one studies on language and texts from the Iberian Peninsula between the Roman Empire and the thirteenth century. Newark: Juan de la Cuesta.

Wright, Roger (2002): A sociophilological study of Late Latin, Turnhout: Brepols.

1 Map 3 represents a slightly revised version of the WALS map of the Perfects (Dahl & Velupillai 2013). See Drinka (2013) for a description of the revisions.

2 Originally middles, the deponents are often identified as passive in morphology but active in meaning, e.g., sequor ‘follow’, morior ‘die’, nascor ‘be born’.

3 ‘The readjustment of the written form to the spoken—and vice versa—is continual […]; the rapid evolution of Latin which, after the break-up of the Romance languages, translates into “errors” of Merovingian and Carolingian Latin, comes incontestably from the influence of the spoken model.’

4 This distribution should not, of course, be regarded as permanent or immutable: Contemporary Standard French, for example, has ceased to use the BE auxiliary productively with newly-coined verbs (Kailuweit 2015: 271); Modern Dutch, conversely, has expanded its use of the BE auxiliary with telecized manner-of-motion verbs and even some transitive verbs (e.g., vergeten ‘forget’)—an innovation facilitated, apparently, by loss of case distinction in Dutch (Gillmann 2015: 354).

5 In these examples, note that periphrastic perfects / passives are bolded, synthetic preterites (Lat. perfects and pluperfects) are underlined, imperfects are italicized, and deponents are italicized & underlined. Translations are drawn, for the most part, from Scholz (1970).

6 The pluperfect of 822, or more accurately the supercompound pluperfect, dispositum habuerit, is comprised of the PPP + pluperfect of habere; the more normal pluperfect of 829, dispositum habebat, is made up of the PPP + imperfect of habere. The supercompound pluperfect formation is enticingly similar to the construction found in Modern French and northern Italian. As indicated by van der Auwera (1998b) and Thieroff (2000) (see Table 1 above), these hyperbolic constructions are distributed across the Charlemagne Sprachbund, and may likewise owe their existence to clerical traditions of Charlemagne’s scriptorium.

7 Thielmann (1885: 549) finds only one example, in the writing of Abbot Valerius (672–81), who wrote during the reign of the Wamba Gothorum: cum iam omnia domus suae ordinata haberet ‘when he had already set everything of his house in order’.

8 The percentages were calculated by using Harre’s (1991: 113; 136) count of haber and haver + PP in historical texts across time, and dividing each by the sum of haber + tener or haver + ter forms in each period, respectively.

9 For example, despite the fact that both Penny (2000) and Tuten (2003) focus on linguistic variation and change in medieval Spanish, neither mentions Arabic as a possible source of structural influence.

10 Albaro’s writing is outside the Carolingian tradition, clinging, instead, to the Visigothic traditions of the past. As Wright notes (1982: 161), Mozarabic writers did not look to the north or to erudite scholars of their own time for models, but to the writings of the Church fathers.

11 Ironically, Albaro’s own progeny provides evidence of this trend: a translation of the Psalms into Arabic was carried out c. 889 by afs̟ b. Albār al-Qūt̟ī, who was, apparently, Albaro’s grandson. It should also be noted that three of the Gospels were translated into Arabic in 946 by Ish̟āq b. Velazquez—a fact seldom acknowledged by scholars (Wasserstein 1991: 6).

12 Following Amman (2002: 322), I use “Standard Arabic” (al-luġatu l-fus̟h̟ā) here as a cover term for “Classical Arabic,” which is closely related to the language of the Qur’ān, and “Modern Standard Arabic”, the present-day supraregional interdialect.

13 Especially frequent in this function are inherently punctual verbs with cognitive, performative, or emotive value. The Perfective is otherwise used primarily for past reference (Cuvalay-Haak 1997: 135–37; Amman 2002: 322); the complex semantic configuration of the Arabic temporal-aspectual system, especially with regard to the perfect, is explored in Fassi Fehri (2003).

14 A more apt parallel would seem to be with those eastern European languages which form their perfects with BE + past active participle.

15 Cuvalay-Haak (1997: 175–96) points out that the aspectual interpretation of the participle, both in Classical Arabic and in the Modern Arabic varieties, is especially dependent on the Aktionsart of the verb, and that in Modern Arabic, it is only telic and momentaneous verbs, not dynamic ones, which tend to have anterior meanings connected with their active participles.

16 The metrical patterns of the xarajāt are hybrid, as well, often setting up a rhythmic contrast between variable quantitative patterns of the muwašša and “Romance-like, fluctuating stress patterns” in the xarjah (Haxen 1991: 46).

17 In his informative chronologization of the development of the muwašša and the xarjah tradition, Corriente (1991: 65–7) notes that the Romance used in the xarajāt may well have been influenced by Arabic metrical constraints, inexact transmission, and an imitation of Andalusian Arabic linguistic patterns.

18 ‘an impoverishment of the system doubtless due to the Arabic adstratum.’

19 The use of BE + PP persists to some extent in archaizing modern literary Portuguese (Mattoso Camara 1972: 145 ftnote).

20 Of 447 examples of the periphrastic perfect in a 17th c. translation of the Old Testament and 62 examples of the New Testament, only 7 and 2 examples of haver + PP are attested, respectively (Suter 1984: 31). Traces of the haver structure still exist in modern literary Portuguese, mostly for archaizing effect, but never in the spoken language (Matoso Camara 1972: 141; Suter 1984: 31–2). The haver construction fell away altogether in the 19th c. (Suter 1984: 216).

21 Squartini / Bertinetto (2000: 410) make the intriguing observation that this restriction applies only to the present form of the periphrastic perfect; other periphrastic forms, such as the periphrastic pluperfect, future, conditional, or subjunctive, are not so constrained. The authors conclude that it is the competition between synthetic preterites and periphrastic perfects which fostered the development of this constraint.

22 In “Lenda de Gaia” by Fernão Lopes, a medieval histiographic work, for example, 159 synthetic preterites appear, but only one periphrastic perfect (Paiva Boléo 1936: 112).

23 Non-iterative uses do exist, especially in some Bible translations of the 17th century, which may represent the textual influence of Spanish or French, or may reflect actual spoken language (Paiva Boléo 1936: 27; 34–5; Squartini / Bertinetto 2000: 419).

24 The published version does not include “mais” ‘more’, but it is supplied in the author’s own handwriting in the signed copy I consulted.

25 In fact, Carballo Calero (1970: 153–4) goes so far as to claim that these perfect forms can be considered “inexistentes” for Galician.

26 See Wright (1994: 265–76) for a thorough discussion of the semantic development of ladino from “the native language (Early Romance) of indigenous Christian inhabitants” throughout the Iberian Peninsula in the 12th century, to “[t]he Romance spoken by non-Christians” by the 15th century.

27 PMC = Poema de mio Cid; EEI = Estoria de España alfonsi; LBA = Libro de buen amor; Zifar = Libro del Caballero Zifar; CORB = El Corbacho; CEL = La Celestina 1499.

28 Rosemeyer (2015: 323) also identifies pockets where ser continues to operate as an auxiliary into the 16th century and beyond, especially with high-frequency “change of location” verbs (e.g., ir ‘go’, venir ‘come’, pasar ‘pass’). This trend is found most clearly in personal letters, where change of location verbs are particularly frequent. Significantly, Modern French likewise preserves the BE auxiliary best with “change of location” verbs in the face of encroachment by the HAVE auxiliary (Kailuweit 2015). Gillmann (2015: 333) designates “change of location” verbs as “core unaccusatives”, from a cross-linguistic perspective.

29 Berschin (1976: 138), in his statistical analysis of La Celestina, found 938 synthetic preterites (76 %) vs. 312 periphrastic perfects (24 %).

30 Alarcos Llorach (1947: 124; 133; 1973: 42–43) also points to the new use of the “presente ampliado” in La Celestina for actions which occurred shortly before the moment of speech. This characterization of anterior function of the “extended present” prefigures and resembles McCoard’s (1978: 123–63) “Extended Now” designation.

31 Brazilian Portuguese likewise retains many archaic patterns, and uses the synthetic preterite more extensively than the periphrastic perfect (Berschin 1976: 44). Lapesa (1968: 358–9) compares this New World preference for the synthetic preterite to that of Galicia, Asturias, León, and Old Castilian. Thibault (2000: 224–25) cautions, however, that innovation has occurred in Spanish on both sides of the Atlantic, and that one should not view Latin American varieties as simply displaying archaic patterns which have not themselves undergone change.

32 The Andean highlands of southern Peru, Bolivia, and northern Argentina show a notably different pattern than other Latin American varieties, using the present perfect much more extensively than elsewhere in South America. Schumacher de Peña (1980: 553–8) claims that substratal influence from the Quechuan of Cuzco and Puno is responsible. Escobar (2000) and Howe (2013) provide extensive evidence for the influence of Quechuan evidentials on the perfects of Peruvian Spanish. The periphrastic perfects of the Spanish of Ecuador have likewise been influenced by the admiratives and evidentials of Quechua (Bustamante 1991: 223).

33 The distribution in the Canaries is currently undergoing change in the direction of Castilian, motivated by sociolinguistic pressures: when the action includes the present, especially with adverbs meaning ‘now’, younger and more educated speakers tend to use the periphrastic perfect more extensively, as in Castilian, and without necessary reference to durativity, while older and less educated speakers tend to retain the conservative pattern. In addition, women are adopting this innovation more frequently than men (Herrera Santana / Medina López 1991: 237; see also Serrano 1995).

34 As Tuttle notes, the prevalence of BE auxiliaries in the east and south of Spain, as opposed to their very limited use in Castilian, may help explain the behavior of isolated varieties such as Capeir, located in the Pyrenees, which has extended the use of BE even further, not unlike some Italian dialects.

35 Umphrey (1913: 18) notes, in fact, that the modern Aragonese varieties which persist in showing agreement are located especially in Upper Aragón, nearest to the French border. Smith (1989: 1989: 322), following Umphrey (1911: 37–39) and Kuhn (1935: 152–55), makes the important observation that such features tend to appear in clusters: those varieties of Aragonese which maintained the BE / HAVE contrast were the same ones which preserved object-participle agreement and which continued to use one lexeme for both the auxiliary and the verb of possession. Smith regards this clustering as signaling a socio-geographically-influenced trend.

36 Menjot (1995) analyzes the provenance of various colonists to the reconquered territory around Murcia in the southeastern Spain, and finds abundant lexical, phonological, and structural mixing due to influence from Arabic, Catalan, Aragonese, and Castilian.

37 The above-mentioned unification in the 15th century of Castile and Aragón, which controlled Catalonia, was clearly responsible for this realignment of cultural and linguistic allegiance. The new Castilian orientation of Catalan is reflected not only in the structural patterns of the language, but also in the language of poetry: the troubadour tradition flourished in Catalan alongside that of Occitan for more than two centuries, but when Castilian became politically ascendant, the language of poetry became Castilian (Paterson 1993: 96).

38 Alongside the HAVE perfects, another periphrastic construction grew up in Languedoc and Catalan and is still in use in modern-day Catalan as the unmarked past tense: the perfet perifràstic, formed especially with the present tense of the auxiliary GO + infinitive. This structure functions as an “extra-heavy foregrounding device” to mark a turning point in a narrative which evokes surprise or special note (Detges 2004: 217–18), or creates “temporal shock” (Mendeloff 1968: 321).

39 Kempas (2014: 85) charts the distribution of aoristic use of the periphrastic perfect across Spanish dialects, especially in hodiernal contexts, and finds that the Basque area appears to be at the forefront of the innovation, both for Spanish monolinguals and Basque-Spanish bilinguals, but that Catalonia is also advanced in this trend, due especially to the obligatory use of the periphrastic perfect for hodiernal reference in Catalan.