RELATIVE PRONOUNS IN LIGHT OF EPIGRAPHIC EVIDENCE: THE PROVINCE OF LUSITANIA *

Summary: This paper focuses on the uses and forms of the relative pronouns as evidenced from the Latin epigraphy in Lusitania. Inscriptions are considered from the 1st to the 8th century AD, with special attention being paid to the future developments in the Portuguese language. To this purpose, other in- scriptions or documents of a different nature dated to later chronologies are also considered as a point of comparison.


INTRODUCTION
Relative pronouns underwent complex changes during the shift from Latin into the Romance languages, which affected phonetics, morphology and syntax. A contrastive analysis of the evidence from antiquity to modern times allows us to track down the progressive development of these forms and to compare previous phases with the current state of Romance languages. In our case, we will focus on the relationship between the Vulgar Latin of the province Lusitania and the modern Portuguese language. It is noteworthy that some of the above-quoted forms are identical with the interrogative pronouns -for instance, que, qual, quanto or onde. This observation can be related with the 'overlapping system' of interrogative and relative pronouns which developed in Vulgar and Late Latin, in consequence of phonetic and morphological 1 'Constituintes relativos ou morfemas-Q'. Cf changes, such as the deletion of final consonants. These processes led to a morphological affinity between separated forms -for instance, between the relative pronoun qui and interrogative-indefinite pronoun quis, both of which converged into Portuguese que. 6 The bonds connecting relative pronouns (Table 2) to their antecedents 7 became weakened already in Vulgar Latin. According to Herman, 'the masculine nominative form qui ("who") and the masculine accusative quem came to be generally available forms that could be used after any antecedent, particularly if they were used to refer to people (…)'; on the other hand, the neutral form quod was often extended to both masculine and feminine antecedents, 'particularly when the nouns are used to refer abstractions or to objects'. 8 Early testimonies of Portuguese are unfortunately scarce, 9 but they show a clear influence of Latin orthography in the spelling of some words. Inscriptions from the 12th and the 13th century were mostly written in Latin (Table 3), reflecting the new status acquired by epigraphy in the Middle Age. Indeed, inscriptions dated to this period are primarily related with the political and cultural elites. In such texts, relative pronouns usually appear in their extended form (without abbreviations) and are clearly deployed according to the norm of Classical Latin. They seem to be completely detached from the changes they had experienced in Vulgar and Late Latin, and come to be even more distant from the Romance system. In such cultivated contexts, Latin was used as an artificial means of expression -often in the frame of poetic compositions -, as a means of favouring linguistic correctness.

Nr.
Date   Table 3. Relative pronouns in medieval inscriptions from Portugal (12th-13th century) 10 In contrast, the principal tendencies of the romance system emerge in full force with the notarial deeds composed in Portugal during the same epoch.
In those examples which are dated to the 13th century, for instance, the form que appears in place of qui, thus anticipating the development into Portuguese que [ke]. 11 Genitive forms such as cuius, cuia and cuium are also recorded within these documents, where they are used as both relative and possessive pronouns; and, besides these, we can also find quanto and qual. 12 However, extravagant uses of the relative pronouns are attested in even earlier documents from Portugal, for example in the Liber Fidei, written in Braga in the 11th century. 13 This text contains several morphosyntactic variants, such as singular quale and plural quales or the merging of que ~ qui in the masculine nominative.

INSCRIPTIONAL EVIDENCE FROM THE PROVINCE LUSITANIA
In the province of Lusitania, whose territory corresponded to a great extent with central and southern Portugal, the modern region of Extremadura and parts of Castile and León and Castile-La Mancha, in Spain, deviant spellings of the relative pronouns tend to be more frequent in inscriptions produced in the Early Period (1st-3rd century AD), while they noticeably decrease in the Late Period (4th-8th century AD). In general terms, relative pronouns are not frequently attested on ancient inscriptions, due to the concise and standardized character of epigraphical texts. In most cases, they are inserted within funerary formulas such as qui / quae uixit annis… or cum quo / cum qua uixit annis…, where they are deployed in the singular forms. 14 An exception to this general trend is represented by the so-called Visigoth Slates, which are documents of a mainly administrative content, dated from the 6th to the 8th century AD, and concentrated in the Spanish sector of our province corresponding to the modern towns of Salamanca, Ávila and Cáceres. Indeed, such documents, whose texts tend to be more elaborated from the syntactical point of view, provide the majority of data referred to the system of the relative pronouns in Late Lusitania recorded in the Computerized Historical Linguistic Database of Latin Inscriptions of the Imperial Age (LLDB), which will be used in this chapter as a main source of information. 15 Acta Ant. Hung. 59, 2019 The Visigoth Slates display, for instance, a clear tendency towards the elimination of the labial trait from the labio-velar phoneme /k w /, as is proven by the frequent spelling COD/COT for quod, 16 as well as by other forms, such as COR for quorum (PizV 92, 1 = LLDB-48264) or CAS for quas (PizV 103, II, 2 = LLDB-59916). 17 It might be observed that this phenomenon -which is to be found, for instance, in Dalmatia and Dacia already in the Early Period -18 seems not to have affected the system of relative pronouns for Early Lusitania. 19 Another evident characteristic of the Visigoth Slates regarding the treatment of the relative pronouns is the devoicing of final -d, as according to the frequent spelling of quod as QVOT (or COT). 20 In the Early Period, such a confusion is well-attested in other provinces, such as Dacia, Dalmatia, Africa Proconsularis and Baetica. 21 We cannot exclude that such a misspelling reflects a general confusion between the relative pronoun quod and the interrogative-indefinite pronoun quot. However, as for the Early Period, it seems more plausible that devoiced spellings of this type were influenced by the phonological context (sandhi) -such in the phrase VOT(VM) QVOT FECIT (IRCPacen 571 = LLDB-16420, AD 1-300) -22 or that they were intended as archaizing forms -such as in the expression QVOT POTVIT PIETAS, as is ---might be explained as technical mistakes or which reflect orthographic habits, such as the use of E for ae, that mainly concerns the pronoun quae, often written as QVE (cf. e.g. LLDB-74822 and 74379). In doing so, we have rejected the possibility that QVE could be a misspelling of the masculine form qui due to the merge of i and e, or, alternatively, that it might correspond to the accusative quem, with loss of final -m. We have also excluded data labelled as fortasse recte and have only used dated inscriptions. Abbreviations of the epigraphic publications quoted can be consulted at LLDB's webpage. 16 Cf. e.g. PizV 4, I, 10 = LLDB-46844 (COD FIAT for quod fiat) or PizV 97, II, 2 = LLDB-59430 (COT ISPENSVM EST for quod expensum est). 17 Venezia 1967Venezia -1968 In Early Lusitania the reduction of the labio-velar phoneme /k w / is attested only before vowels such as a or i, as is proved by spellings such as QADRATI for Quadrati . 5745,7568,9146,9791,9875,9878,9886,9887 (Early Dacia); LLDB-266, 14294, 34870 (Early Dalmatia);LLDB-43658, 50624, 51126, 54273, 65286, 82231 (Early Africa Proconsularis) and LLDB-22112, 25096, 26080, 74374 (Early Baetica). 22  recorded in a verse inscription (ERAEmerita 140 = LLDB-60398, AD 151-200), i.e. in a text with literary connotations. 23 In the Late Period, the oscillation between -t and -d might reflect the instability of the final consonants, as has been correctly pointed out by Velázquez Soriano. 24 Indeed, in the province of Lusitania -but also in the Baetica -the loss of final -d is effectively attested with reference to the relative pronouns already in the Early Period. 25 However, it is also possible that, in those cases where -d is absent, we are dealing with a phenomenon not of a phonological but of a syntactical nature, that is, the confusion between the ablative (quo) and the nominative-accusative (quod). Examples of case merging in the domain of the relative pronouns are in fact attested in both Early and Late Lusitania. As for the Late Period, they mainly focus -once again -on the Visigoth Slates, 26  , which shows the interchange between the dative and genitive case, which is a typical phenomenon of Late Latin. 27 In general terms, the Visigoth Slates demonstrate a sort of 'fixed use' of the neuter pronoun quod, 28 which can alternatively replace a feminine form, such as in 29 COT ISPENSAS SVNT for notitia ueruellarum quae expensae sunt (PizV 97, I, 2 = LLDB-59415), or which can even refer to a plural name, such as in NOTITIA DE CEVARIA QVOT DEVES for notitia cibariorum quae debes (PizV 52, I, 3 = LLDB-48073). 30 Conversely, the overlap of the masculine as the 'universal gender' in the domain of the relative pronouns is scarcely attested in Lusitania. On a private letter dated to the 3rd century AD that was rudely scratched on a shingle clay in Emerita Augusta (Mérida), we read the sentence PVELLAM QVI IAM FETO TOLLERAT for puellam quae iam fetum tulerat (Mallon p. 550 = LLDB-45890), where the nominative masculine pronoun qui is used to refer back to the feminine noun puella. In later times, on a paleo-Christian verse epitaph from Portugal, we find the constructions ILLA (…) QVEM for illa (...) quae and ME MISERAM QVI for me miseram quae (CIPTP 3 = LLDB-29903 and 29915). In other provinces, this phenomenon is instead more widely reported, both in the Early Period -particularly in Africa Proconsularis and Dalmatia -and in the Late Period -especially in Gallia Narbonensis, Venetia et Histria and Dalmatia. 31

CONCLUSIONS
In conclusion, epigraphic materials from different periods allow us to reconstruct the evolutionary process that led to the progressive disintegration of the Classical Latin system of the relative pronouns. The first hints of such an evolution are visible in the documents of the province of Lusitania and date to the Roman age and Late Antiquity. The subsequent adoption of Latin as a cultivated language and the elitist use of inscriptions in medieval times led to a sort of 're-normalization' of the epigraphic language, including the system of the relative pronouns. Nevertheless, complementary documents of a different nature, such as non-epigraphic texts from the Middle Ages, provide us with a wider panorama over the phenomena of linguistic change. By the 11th century AD, the system of relative pronouns of the Portuguese language had already taken its basic shape. Open Access. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited, a link to the CC License is provided, and changes -if any -are indicated. (SID_1)