Fronting in Old Spanish

The article focuses on a cross-genre study of fronting phenomena in 13th Century Old Spanish. In particular, we study stage topics and deictic fronting, fronting in quotative inversion, quantifier fronting and information/broad/weak focus fronting comparatively, for the first time, in three types of texts (that is, (a) Cantar de Mío Cid, (b) La Fazienda de Ultra Mar, and (c) Estoria de España and General Estoria) while we compare with equivalent judgements from Modern Spanish in order to establish what exactly has changed. “Fronting” a term which has been used to describe all sorts of configurations ranging from stylistic fronting to focus fronting to non-focus fronting receives here a principled discussion. We show that, overall, fronting with a verum focus interpretation has largely been preserved into Modern Spanish, albeit often restricted, while the most notable change seems to be the loss of a preverbal focus position conveying broad focus. In doing so, we reconcile Leonetti’s (2017) claims for an informational partition, which does not divide the fronted element from the verb, but, rather, the fronted element together with the verb from the postverbal subject with our own claims about the syntactic mechanisms which yield this partition. We proceed to conclude that fronting operations are not a derivative of the V2 parameter being operative (contra Wolfe 2015).


Introduction
In this paper, we examine three types of Old Spanish texts and compare data involving fronting against Modern Spanish native judgements in order to find out to what extent the different types of fronting in Old Spanish texts are still possible in Modern Spanish.
Fronting corresponds to different syntactic configurations (equally involving A'-dependencies and base-generation) and various discourse readings (sub-types of both topic and focus), namely: stylistic fronting (see Elvira 2018), fronted stage topics (either locative or temporal), deictic fronting (see Martins 2010, 2011), fronting in quotative inversion (see Leonetti 2017: 896), quantifier fronting, scrambling, clitic left dislocation (hereafter CLLD, see Bouzouita 2014; and focus fronting be it contrastive or informational (see Sitaridou 2011, Batllori andHernanz 2015, a.o.). Although some authors use the term "fronting" as synonymous to focus fronting operations (see Cruschina 2008 et seq.; in fact, Cruschina uses the term exclusively for information focus, not for contrastive focus), others use it to suggest a type of non-focal fronting (see Leonetti 2017). We follow Biber et al (1999: 900) in that "fronting refers to the initial placement of core elements which are normally found in post-verbal position". It follows that, for us, fronting is a theory-neuter term to capture different types of preposing and it is precisely this "informal vagueness" of the term (see Cormack and Neil 2000: 387), which partly motivates the present paper. To delimit the discussion on fronting types, we focus here on Glossa general linguistics a journal of Batllori, Montserrat and Ioanna Sitaridou. 2020. Fronting in Old Spanish. Glossa: a journal of general linguistics 5(1): 61. 1-39. DOI: https://doi.org/10. 5334/gjgl.893 stage topics, deictic fronting, fronting in quotative inversion, information/broad/weak 1 focus (to be defined), and quantifier fronting, most of which involve verb-subject inversion and strict adjacency between the fronted element and the verb.
Fronting in Old Spanish is linked to a number of key issues; in particular: (a) to what extent it can be associated with pragmatic neutrality (namely, the commonest, canonical or unmarked word order, to the extent of course such a thing does exist) or lack thereof, which would, in turn, feed into a non-V2 analysis of Old Spanish given that it will offer evidence against formal movement (see Sitaridou 2019 for the most updated summary of the V2 debate and arguments against a V2 analysis; see Wolfe 2015 for fronting operations as a derivative of the V2 parameter being operative; see Hsu 2017 on how OVS orders are a case of so-called "relaxed V2 system"); (b) the role of genre in triggering word order variation, that is, whether or not some texts exhibit certain fronting possibilities; and (c) establishing whether fronting is preserved in Modern Spanish or not, thus affording us some insights into diachrony.
The structure of the paper is as follows. In §1.1 we discuss the choice of texts, the problem (and opportunity) posed by genres as well as the methodology we have used; in §1.2 we present the fronting puzzle and some relevant aspects of Leonetti's (2017) analysis on which we shall base parts of our discussion; and in §1.3 we present our main hypotheses to test and claims. The discussion of the data starts with stage topics and deictic fronting in §2 while quotative inversion, quantifier fronting and information/broad/weak focus fronting are discussed in §3, §4 and §5, respectively. In §6 we summarise our findings and discuss some diachronic implications following from the discussion of the data in this paper.

Texts, genres and methodology
In this article, we examine fronting possibilities in three types of Old Spanish texts. The reason why we decided to compare three different types of texts not only does it relate to the perennial issue of representativeness of 13th Century speech, but, rather, it is a conscientious effort to be mindful of the constraints, coding norms and traditions of medieval writing (see Schneider 2008). Syntax-discourse phenomena such as fronting are particularly susceptible to discourse traditions: for instance, consider the role of latinising syntax in triggering grammaticalisation (see Bouzouita 2019;Cornillie & Octavio de Toledo 2015) and the broader discussion on syntactic variation and genre (Dorgeloh & Wanner 2010); hence the need to establish which fronting configurations are due to normative bias and/or discourse traditions or, rather, reflect core, genre-free syntactic mechanisms.
Our choice of texts is guided by Badia i Margarit's (1960) and Cano's (2006) comments concerning the construction of discourse in the history of Spanish. Both authors consider (i) a. Modern Spanish Mi abuela dice que mucha chorrada hace el Gobierno. my grandma say.prs.ind.3sg that a.lot silly.things do.prs.ind.3sg the government 'My grandma says that the government keeps buggering about a lot.' (Gallego 2007: 221;Torrego 1980) (ii) b. Modern Portuguese Muitos livros os meus pais me deram. many books the my parents dat.2sg give.prf.3pl 'Many books did may parents give me' (Raposo and Uriagereka 2005: 691, footnote 19) As is well-known, Uriagereka (1995) states that the examples under study in his work do not pattern with a V2 analysis and considers this type of constituents to be in a sentence internal 'F category which encodes point of view' (Uriagereka 1995: 163, 169). Ambar (1999) considers them as evaluative expressions. epic poetry such as Cantar de Mío Cid (thereafter Mío Cid) 2 to be closer to the oral organisation of discourse than Alfonso X's Estoria de España and General Estoria (thereafter GE&EE) 3 prose. According to them, Mío Cid was written in order to be recited/narrated to a wider audience, whereas Alfonso X's prose was written to be read to a learned, smaller audience; hence, in the latter, every concept needs to be specified, elaborated and clarified by means of abundant subordination for the purpose of which, a wide range of innovative conjunctions/subordinators is introduced for the first time (see Elvira 2002). On the contrary, as shown by Batllori and Suñer (2008), in Mío Cid, embedded clauses display oral strategies 4 significantly more than in Alfonso X's prose. Moreover, to further complement our selection of genres, we have also included La Fazienda de Ultra Mar (thereafter Fazienda) 5 because, according to Sanchis Calvo (1991), its prose reflects the koiné 6 of the time. 7 To source data from the above-mentioned texts we have used CORDE (i.e., Real Academia Española Corpus Diacrónico del Español) following some necessary modifications in the search function, for instance: we specified the search for Autor 'author' and Obra 'piece of work'. Hence, when examining the data of Mío Cid, we inserted Anónimo in the slot provided for Autor and Poema de Mio Cid in the one for Obra; for the data extracted from Fazienda, 8 we put Almerich for Autor and La fazienda de Ultra Mar for Obra. Finally, for GE&EE, we wrote Alfonso X in the Autor slot and either General Estoria or Estoria de Espanna (or Estoria de España, depending on the edition we wanted to access) in the Obra one. Moreover, we contrasted the examples obtained with other editions of the same works. 9 As for the Modern Spanish data, some examples were extracted from CREA (i.e., RAE, Corpus de Referencia del Español Actual) or provided by the first author. 2 A 12 th Century epic poem. 3 A 13 th Century prose. 4 For instance, consider the copious use of que 'that' for both coordination and juxtaposition as well as for different types of embedded clauses (such as causal, purpose, concessive and conditional ones, among others) in combination with different verbal moods (indicative vs. subjunctive) and word order. 5 An early 13 th Century itinerary with some parts translated from the Bible. 6 The suspected Gallo-Romance origin of the translator has been adduced to explain the abundance of Aragonese, Catalan or Occitan lexical and syntactic features (apart from the Hebrew syntactic loans attributable to the original text). However, in recent years the Gallo-Romance origin of the translator has been brought into question and has now generally been accepted that he could have indeed been from the Iberian Peninsula: 'presenta fuertes caracteres aragoneses, por lo que tal vez habría que situar su redacción en la zona fronteriza entre Castilla, Navarra y Aragón, a falta de datos precisos' (it presents a strong Aragonese character, which is why we could possibly situate its redaction in the border area between Castille, Navarre and Aragon) [translation our own] (Rodríguez Molina 2010: 747). This resonates with Sanchis Calvo (1991: 550), according to whom, 'rasgos semejantes a los mencionados se encuentran en textos castellanos de fines del XII y de la primera mitad del XIII, época de fuerte influencia ultrapirenaica en España' (features similar to the ones mentioned are found in Castilian texts at the end of 12th Century and the first half of 13th Century which is an era marked by strong influence from across the Pyrenees in Spain) [translation our own]. In fact, Sanchis Calvo's dissertation was supervised by Lapesa, who had labeled the influence of oriental features (Aragonese, Catalan, Occitan, etc.) in the charters he had studied, as part of the Spanish koiné of that period. 7 One reviewer correctly points out that Fazienda contains parts of the Bible, which in itself is a collection of different text types (Rosemeyer and Enrique-Arias 2018; see also Vincis 2008;2009, etc.). Therefore, the reviewer rightly suggests we should verify from which book of the Bible each fronting example comes from in order to determine its genre (e.g., lyrical, sapiential, narrative, prophetic, etc.). This is, in fact, the reason why the discourses contained in Fazienda and GE partly overlap. Crucially, this issue has not been taken into account in this paper -we leave this task to future work. 8 Notice that CORDE contains Moshé Lazar's [Universidad de Salamanca (Salamanca), 1965] edition of this work. We know that this edition has been recently described as riddled with interpretational errors (see Rodríguez Molina/Octavio Toledo y Huerta 2017: 10), but it is the edition chosen by RAE and we therefore, use it for methodological coherence; however, we scrutinized each case for any potential philological problems. 9 For instance, in the case of Mío Cid, CORDE contains the edition by Montaner [Crítica (Barcelona), 1993]. We contrasted the data of Montaner's edition with Menéndez Pidal's paleographic edition [Biblioteca Nacional (Madrid), 1961] and eventually adopted the verse numbering of the latter because CORDE does not include any numbering. As for Fazienda, we contrasted the data extracted from CORDE, whose edition is Lazar's [Universidad de Salamanca (Salamanca), 1965], with the actual book and we took the page numbering from the latter. With regards to General Estoria and Estoria de Espanna que fizo el muy noble rey don Alfonsso (or Estoria de España, which corresponds to different parts of Alfonso X's work), the data from the former are extracted from CORDE as well as the data of the Estoria de Espanna, which correspond to the The data sourced from the above-mentioned texts have not been treated statistically. In this sense, we agree with Leonetti (2017: 889) in that textual frequency "is not always a reliable criterion, since frequency may vary from one type of text to another" and, furthermore, we add, it is an even less reliable a criterion given that frequency can vary from one part (i.e., the prologue) to other parts of the same text.

The fronting puzzle
According to Leonetti's (2017) approach to word order, dislocations and focus fronting aside, there is an additional fronting mechanism which yields X/OVS in both Modern and Old Spanish, whereby, the fronted element does not seem to bear any focus hence why Leonetti dubs this fronting operation non-focal fronting -consider (1) as exemplified from Modern Spanish: 10 (1) Modern Spanish (Leonetti 2017: 910) Más alumnos teníamos el año pasado. 10 more students have.ipfv.ind.1pl the year pass.ptcp.m.sg 'We had more students last year.' As such, he introduces an informational partition, which does not divide the fronted element from the verb, but, rather, the fronted element together with the verb from the postverbal subject. We agree that in examples such as (1) and (2) there cannot be any intonational breaks following the fronted material or emphatic stress on más alumnos 'more students'. Importantly, however, (2a) is not a felicitous answer to a 'What happened?' type of question, to which the felicitous answer would be (2b), instead; crucially, informational/broad focus does not trigger fronting in Modern Spanish. 11 (2) a. Modern Spanish (Leonetti and Escandell-Vidal 2009 Importantly, for Leonetti, the interpretation of the fronted constituent in (2a) is underspecified and context dependent: it could be a topic, wide/informational focus, or devoid of any specific informational load. The latter, whereby the fronted material does not have any specific informational load, would have important ramifications for the V2 debate since it could be used as evidence for claiming that the fronted material is the result of formal movement, not A'-movement (see discussion in §2 and Sitaridou 2015; 2019). parts of these works edited by Sánchez Prieto while the latter (i.e., Estoria de España) follows the edition of Kasten andNitti (1978/1997). 10 The glossing system followed in this paper is consistent with the conventions for interlinear morpheme-bymorpheme glosses provided by the Leipzig Glossing Rules. Examples unless otherwise indicated are from Old Spanish.
In the examples we use italics to indicate non-contrastive fronting (especially when we refer to examples from Leonetti 2017); bolding for subjects; and small caps for focus. 11 To illustrate a context in which (2a) can be produced, consider the following scenario: speaker A is disturbing speaker B, who has got a lot of work to do. In this context, speaker B would then say: "Déjame en paz. Bastante trabajo tengo hoy, como para estar aguantándote todo el día." [Leave me in peace. I have got enough work today to be putting up with you as well.] Thus, (2a) is a discourse-linked utterance, whereby 'enough work' is not focused, yet it highlights the fact that speaker B has enough work. In terms of prosody, this type of construction has a different prosody from unmarked asserted sentences (a particular intonation, in fact). However, it is completely different from that of contrastive focused constituents.
Although Leonetti does not offer any technical implementation, his proposal is truly minimalist in spirit since it seems to rule out features like [focus] and/or [topic] from being present in the derivation. Nevertheless, at this point there are a few crucial questions with Leonetti's analysis: (a) To what extent is this proposal tenable from a theoretical perspective given the well-known quantificational properties of focus versus topics; in other words, can topics and foci occupy the same position? (b) Even if we do not assume a cartography framework, but a minimalist one instead with fewer or even a single CP projection, which could presumably be occupied by both topics and foci, how would that work in terms of phases given that the verb is in T 0 (a weak one) and the fronted material is in spec-CP (a strong one) assuming, as we shall show, adjacency? (c) How are we going to explain the fact that although topics and foci occupy the same spec-CP position they trigger different configurations, namely enclisis and proclisis for pronominal object pronouns, 12 respectively?

Hypotheses to test and main claims
Fronting, given the discourse-syntax nature of the phenomenon, is independently known to be vulnerable to attrition and change (see Tsimpli et al. 2004). As such, it allows us to test for the following hypotheses: a) If structural V2 is operative, we should not find any register variation affecting fronting possibilities; in other words, fronting should operate under identical structural conditions irrespective of register differences; b) Fronting possibilities in Old Spanish should be fundamentally different from Modern Spanish ones, if the former was V2 and the latter is no longer so; c) Non-focal fronting is an operation to "avoid focus" (as claimed for scrambling/interpolation by Martins (2002) for Old Portuguese) or it can actually convey information/broad/weak focus while still distinct from quantificational contrastive focus; d) Contrastive focal fronting should have a different grammar from non-focal fronting given that the former is quantificational while the latter is not.
It is argued that fronting operations in 13th Century Spanish and the organisation of Information Structure (hereafter IS) in general, in spite of some register variation, are intrinsically similar to Modern Spanish. 13 Any differences can be attributed to rhetorical restrictions and schemata, which the Medieval Ars dictaminis and Ars dictandi imposed on writing (see Batllori 2015;Sitaridou 2015;Elvira 2017). The single most important difference in the organisation of IS remains the unavailability of fronting objects and other non-finite verb forms -both of which are linked to an OV syntax of the past, namely the Latin one. Our findings also show that (a) in neither Old nor Modern Spanish all stage topics are in spec-TP, but that some may indeed be generated in the left periphery; (b) deictic fronting is neither focus nor topic: while it refers to something previously mentioned, it is used in order to elevate the referent to a higher level of salience, thus it cannot be said to be entirely non-focal; (c) fronting in quotative inversion has remained stable; (d) only one type of quantifier fronting has survived whilst it got grammaticalised. Overall, our findings offer further support to the idea that Old Spanish declaratives exhibit an unmarked SVO order (Sitaridou 2011;Dufter and Octavio de Toledo 2014;Castillo Lluch and López Izquierdo 2015, and the works therein) and fronting, including OVS, should be considered an expression of different IS values (Sitaridou 2011;Eide and Sitaridou 2014;and Leonetti 2017) rather than V2.

Stage topics and deictic fronting in Old Spanish
In Modern Spanish, as well as in several other languages, fronting of temporal and locative expressions -(3) and (4), respectively -anchors the meaning of the sentence in a particular time or place of the narration providing a scene setting topic value (see Erteschik-Shir 1997), and it is usually produced in the absence of a canonical "aboutness topic" (in the sense of Frascarelli & Hinterhölzl 2007). Ojea (2017) argues that in Modern Spanish (3)-(4), the discourse legible features of these expressions are edge features that are targeted by T and thus occupy spec-TP. 14 (3) Modern Spanish (Ojea 2017: 80) Hoy llegaron los turistas franceses. today arrive.prf.3pl the tourists French 'The French tourists arrived today.'

(4)
Modern Spanish (Ojea 2017: 80) Aquí falta el catálogo de la exposición. here lack.prs.ind.3sg the catalogue of the exhibition 'The exhibition catalogue is missing here.' Fronting of temporal and locative expressions in Old Spanish was extremely frequent (see Mackenzie & van der Wurf 2012), so there is no doubt as to whether this type of fronting was operative or not. What remains debatable is the sort of movement it involves (formal movement or A'-movement) as well as the landing site (spec-TP or in the left periphery). To this end, clitic placement may provide a useful diagnostic since it has been shown to correlate with IS. In particular, it is trivially considered that focus triggers proclisis, whereas topic correlates with enclisis (see Bouzouita 2008;Fernández-Ordóñez 2008-9 for Spanish, among others; Costa & Martins 2011 for European Portuguese;Donaldson 2016, for Old Occitan). In light of this, consider (5): (5) (Fazienda: 101) E agora dyovos el Criador tierra & now give.prf.3sg=dat.2pl the Creator land 'and now the Creator gave you land' Example (5) shows enclisis, which would then have to be taken to indicate that the stage topic agora 'now' is not in the left periphery. 15 However, by far the commonest order is proclisis; in fact, temporal and locative adverbs attested in Mío Cid, in (6) and (7), trigger proclisis (as shown by 6a, 6b, 6g, 7a, 7c, 7e, 7h, and 7j). Notice that in (6a) and (6b), as well as in (7a) to (7e), the subject is postverbal and there is strict adjacency with the verb (except for 7a), which may well be a case of right dislocation 16 . In contrast, in (6c), (6d), (7f), and (7g) the subject is a left-peripheral topic (as indicated by the fact that the subject also precedes the adverb) and the reading of these sentences is not thetic but categorical (see also Granberg 1988;Bouzouita 2008a;b Aquel poyo, en él priso posada that stone.bench in it take.prf.3sg rest 'On that stone bench, he rested on' Provided that the correlation between proclisis and focus is correct, fronted expressions in (6)-(7) are either (i) not true stage topics (as defined by Leonetti 2017) or (ii) occupy a left-peripheral position in Old Spanish. On the one hand, for Leonetti (2017), as previously discussed, these preverbal XPs are unaccented and form one single unit alongside the verb while bearing no focus. On the other hand, Costa and Martins (2011) consider deictic fronting (which, according to them, includes fronting of deictic adverbs -i.e., temporal, locative and manner ones such as the ones in (6) and (7)) to convey contrastive focus. We do not agree with either analysis because we believe that there is a third alternative, which in fact captures insights from both approaches, namely: deictic fronting involves movement to the left periphery as indicated by proclisis whilst the interpretative value of salience could be viewed as akin to broad/informational focus. 18 The discussion at this stage is very reminiscent of Speyer's (2010) discussion about the loss of unaccented fronting in the history of English. According to him, who essentially follows Frey (2004Frey ( , 2006, there are two possible fronting mechanisms in the history of English: (a) True A-Bar Movement (TAB), which, in turn, results in a contrastive interpretation on the fronted XP; or (b) Formal Movement (FM), which has no interpretive effect. Clearly, the latter is fully compatible with V2 syntax. In Sitaridou (2015) there was extensive discussion of whether FM can be detected in the fronting of participles and the answer was a negative. Before we return to these issues, let us now explore the examples of deictic fronting farther.
Moreover, there are other types of deictic expressions (mostly demonstratives, manner expressions or identity adjectives) that also help to set out the discourse background and which are often dubbed as deictic fronting (see Costa and Martins 2010  In Mío Cid, we observe that (a) with regards to clitic placement we find mostly proclisis (8a, 8b, 8e, 8g, 8h and 8i); (b) the occurrence of deictic adverbs triggers verbsubject 19 inversion (see also Meyer-Hermann 1988) -as shown in (8a) and (8b), and in (7b) to (7e), but not for the cases in which the subject is a base-generated left-peripheral topic (8e and 8f). Notice that left-peripheral subjects are also compatible with a fronted temporal or locative adverb (6c, 6d, 7f, 20 and 7g), which renders them compatible should the latter be focused. In line with Donaldson (2016), for Old Occitan, we consider that proclisis ensues when (i) spec-FocP is filled or (ii) a constituent has moved to a left-peripheral position. Even in the cases of clitic left dislocation (CLLD) contexts, if there is fronting of a focal constituent (i.e., assí 'thus'), there is proclisis and we consider the clitic to have moved at the left of Vᵒ, namely in Tᵒ.
Therefore, for us, it is not the case that all fronted temporal, locative, manner and deictic expressions are necessarily stage topics in spec-TP. Some of this fronted material undergoes movement, the interpretational result of which is a new information focus reading. This can be seen in (8c), for example, which could be envisaged as the answer to What happened? The context is the following: the Cid and his daughters were crying. Then, one could ask and what happened? and the answer could perfectly be So did the knights.
As for the IS value of the subject, let us start with Modern Spanish: VS triggers narrow focus on the subject (or wide focus with unergatives) whereas SV expresses a categorical reading with a topic-comment partition in unaccusative configurations (following Leonetti 2017: 895). Incidentally, most often, when the subject is inverted or absent, the verb occupies the second position. However, this cannot be taken as evidence for a prototypical V2 configuration because many of the examples showing inversion exhibit, in fact, V3 -as in (6c), (6d), (7f), (7g), (8e), (8f), (8g), (8i), (8k) (see Sitaridou 2019 for a detailed discussion on V3 constructions).
So far we have shown the existence of deictic fronting in the data of Mío Cid, which is the Old Spanish text in which fronting seems to reach maximal exponence (and consequently, a text with a high frequency of linear V2). Martínez-Gil (1989: 896 and1898) and Elvira (2017: 179) attribute this aspect of the text to the prosodic requirements imposed by the meter and the rhythm of the composition. Prosody and, in particular, the use of the colon 21 (see Adams 1994 for the same concept in Latin) yields a composition with plentiful fronting, the value of which is more akin to informative than focal, always according to the above-mentioned authors. Next, we provide evidence for the existence of similar strategies in the other Old Spanish texts under study.
As for Fazienda, our prediction is that we will also find cases of fronting of temporal, locative, manner adverbs and deictic expressions independently of the different rhetoric nature of the text. The important thing to look out for, though, is whether fronting triggers proclisis or not. That is, whether the fronted material is going to be in spec-TP or in a left-peripheral focal projection, like in Mío Cid. Notice that in Fazienda temporal and locative adverb fronting are mostly used with unaccusative verbs like ir 'go' -in (9a) and (9b)-, venir 'come' -in (9c)-, but they can also be used in other unaccusative configurations (for instance, in passives, in (9d), or locatives, in (9e)) or even with other types of verbs, like in (9f) and (9g). In general, these fronted deictic expressions display postverbal subjects; however, there is no need for strict verb-subject adjacency, as illustrated in (9b). Crucially, (9g)[=(5)] shows that under some circumstances this type of preposing could correspond to a stage topic because it does not involve movement to the left periphery and does not trigger proclisis. Hence, Fazienda might have stage topics in spec-TP.  As for deictic fronting in Fazienda, the data from (10a) to (10b) and (10f) show that it triggers proclisis and (10d) to (10f) illustrate strict verb-subject adjacency (in particular, (10d) shows that, even in the case of passive structures, the subject is adjacent to the auxiliary verb). Let us turn our attention now to GE&EE in which fronting of temporal, locative, manner adverbial expressions and deictic ones (11) is also widely attested (see also Eide and Sitaridou 2014) with no noticeable difference from the other two texts in terms of distribution and discourse value for either the fronted material or postverbal subjects: 22 (11) a. Regarding the preposing of locative adverbs with unaccusative verbs and existential estar 'be' (from 11a to 11c), the examples of GE&EE seem to pattern with Modern Spanish, except for the fact that the subject is not always adjacent to the verb. These cases might be analysed the same way as Modern Spanish stage topics (in Spec-TP position). However, (11d) to (11f) display proclisis and also the verb of the former is unaccusative and its subject is postverbal and adjacent to it. (11g) is one of the many examples of V4 that can be found in these works. As for (11h), the verb is unaccusative and the subject is postverbal and introduced by the conjunction que (in a clear case of recomplementation), and assi 'thus' might also be analysed as a stage topic. However, in (11i) there must have been movement to the left periphery because we have proclisis. 23 Finally, it can be seen that the preposing of the deictic expression that contains a demonstrative also triggers proclisis (11i). Hence, even if we could think that some sentences could be considered as examples of stage topics in spec-TP, there are others that involve movement to the left periphery.
To sum up, the only register variation found in the 13 th C. texts that we have examined relates to the fact that in Mío Cid preverbal temporal, locative, manner adverbs and other deictic expressions move to the left periphery and we may say that even if they are not focal they may convey information focus, which could be checked in a position external to FinP/TP because it triggers proclisis. 23 Consider also cases where the adverbs do not move to the left periphery, as in (i)  assí dezimos nós: a lo que demanda el Cid … so say.prs.ind.1pl we to the that ask.prs.ind.3sg the Cid '(If it pleases our king) so we say: "Whatever the Cid asks (you help him with)"' Apart from temporal, locative and manner adverbs being fronted, quotative inversion is also documented with fronting of objects: the direct object, as in (13a), the indirect object, as in (13b), or both, as in (13c), and even a fronted embedded clause, as in (13d). The organisation of IS in quotative inversion in Mío Cid works as follows: (a) when there is a temporal, locative or manner adverbial expression, there is narrow focus on the subject (12). Notice that this has remained stable to present day (14a). In contrast, in OVS configurations, which are illustrated in (13), fronting triggers subject inversion, but it does not convey narrow focus on the subject. We agree with Leonetti (2017: 907-911) in that the fronted constituent can neither be interpreted as a topic nor as a contrastive focus and also in that fronting contributes to IS in an indirect way-since there are not intonational breaks-because the fronted element can create a connection with a discourse antecedent -like in (13a) -, without however being an aboutness topic. This use has also been preserved in Modern Spanish (see 14b). Besides, in most cases it gives rise to verum focus (see 13c, for instance). Esto dijo the death all =acc.n.3sg equal.prs.ind.3sg this say.prf.3sg la afable María Gloria … the affable Mary Glory 'The affable Mary Glory said this: "Death is the great equaliser"' As for Fazienda, XVS also denotes narrow focus on the subject, as illustrated in (15) Assi dize el Criador: con estos descornaras a los so say.prs.ind.3sg the Creator with these dehorn.fut.2sg to the de … of 'So says the Creator: "With these you will dehorn the ones of (Syria)"' d.  (18a) or they are narrations of biblical stories, as shown in (18b-d). Notice that in (18a) the clitic pronoun is proclitic, which implies that there has been movement of the temporal deictic adverb to the left periphery. Thus, in quotative inversion contexts, XVS patterns with Mío Cid and OVS is configured the same way as in Mio Cid and Fazienda. Moreover, in GE&EE, VS is also the most frequent context to convey narrow focus on the subject-see (18e) and (18f) Overall, we note that quotative inversion seems to have remained stable in Modern Spanish (but for the cases in which we have proclisis, such as (18a), which have certain epic flavour). For the rest, a similar pattern is found, consider (19)

Quantifier fronting
According to most studies on quantifier fronting (QP fronting hereafter) in Modern Romance (Ambar 1999;Quer 2002;Leonetti and Escandell-Vidal 2009;Escandell-Vidal and Leonetti 2011;Cruschina and Remberger 2017, among others), QP fronting does not involve any kind of informational partition since it conveys an evaluative value (see Ambar 1999) that Leonetti (2017: 913) identifies with a verum focus interpretation denoting narrow focus on the positive polarity of the utterance and thus bringing about the overall strengthening of the assertion.
In Old Spanish and, in particular, in Mío Cid we can identify two main types of QP fronting. In the first type, which is still grammatical in Modern Spanish, the fronted quantifier modifies the whole sentence -see (20) -and fronting gives rise to a verum focus reading (see Leonetti and Escandell-Vidal 2009 for verum focus in Modern Spanish). Also, as illustrated in (20f) In the second type of QP fronting, the preposed quantifier modifies the predicate of a copulative sentence with ser 'to be' (21a,b) or the definite direct object of a light verb construction with auer 'to have'-see (21c). The sentences in (21) are all ungrammatical in Modern Spanish (22). As shown in (22), (21a) and (21b) would be expressed by means of muy 'very' nowadays, whereas (22c) would need a subject experiencer verb because the arguments of Old Spanish auer 25 denoted alienable possession, but this value is no longer available with haber in Modern Spanish. In the latter cases, the construction conveyed the superlative quantification of the alienable abstract noun (of emotion or affection); thus, apparently, no informational partition was created through QP fronting in these structures. Notice, though, that, in some way, the examples in (21) could be interpreted as Modern Spanish clefts, as for instance in (22e) to (22i). If so, we would then have to say that there is a focus-background partition with contrastive narrow focus on the constituent that precedes the verb ser 'to be'. Finally, there are other structures involving QP fronting in which the prepositional phrase (PP) that modifies the quantifier is extraposed -see (23) -but in these cases either the quantifier has to be associated with que 'that' and therefore functions as a correlative consecutive embedded clause conjunction -as in (23a,b) -or the juxtaposition of two clauses with QP fronting in the second clause yields the interpretation of the logical causeconsequence relation between the two, as in (23c).
(23) a. (Cid: v. 2529) tantos avemos de averes que no son so.many have.prs.ind.1pl of possessions that not be.prs.ind.3pl contados count.ptcp.m.pl 'We have so many possessions that they cannot be counted' b. (Cid,v. 785) Tantos moros yazen muertos que pocos bivos so.many Moors lie.prs.ind.3pl dead that few alive á dexados have.prs.ind.3sg leave.ptcp.m.pl 'There were so many Moors lying dead that he has left just a few alive' c. (Cid: v. 1800) Alegres son por Valencia las yentes cristianas, ¡tantos happy be.prs.ind.3pl in Valencia the people Christians so.many avién de averes, de cavallos e de armas! have.ipfv.ind.3pl of possessions of horses & of weapons 'The Christians were happy in Valencia. So many goods, and horses and weapons they had!' Interestingly, the only type of QP fronting attested in Fazienda (scarce though), is the first type (i.e., the one still available in Modern Spanish)-see (24a). The second type of quantification for which we would use muy 'very' nowadays is also attested, but in this case the quantifier appears before or after the predicate it modifiessee (24b) and (24c), respectively. Otherwise, sentence-final position of the quantifier yields quantification over the whole sentence, as in (24d). In (24c) it quantifies the clause fue albergada and in (24d), the purpose clause fer bevir pueblo 'make people live'. ind.3sg to make.inf live.inf people mucho much 'And you inflicted me evil and God turned it into good, as is today so as to make people live more' In GE&EE, the two types of QP fronting are also widely attested. On the one hand, we find the type of QP fronting that triggers a verum focus reading, as in (25), which is still available in Modern Spanish. On the other hand, Alfonso X's GE&EE still exhibit some examples of the kind of QP fronting that modifies the predicate of a copulative sentence. As mentioned above, this could be related to the fact that in these works some epic poems (such as the Mío Cid and the Poema de Fernán González, for instance) were turned into prose. In contemporary Spanish (26a) to (26e) would be ungrammatical, but (26e) and (26f) could be analysed as examples of cleft sentences with the adverb muy instead of mucho -see Camus (2008)  Mucho espantado fue Jacob d'aquel fecho que fizieron a.lot scare.ptcp.m.sg be.prf.3sg Jacob of.that deed that do.prf.3pl 'Jacob was really scared about that action that they took' As previously seen, this second type of QP fronting might have expressed an informational focus-background partition in Old Spanish comparable to the one exhibited by the examples in (27f) and (27g), which might be given the same value as Modern Spanish clefts and thus denote contrastive narrow focus on the fronted constituent.

Information/broad/weak focus fronting
Information/broad/weak focus fronting (see Batllori and Hernanz 2015) 26 involves objects (27), adjuncts (28), adjectives (29), participles (30), infinitives (31), gerunds (32), and adverbs (33), all of which give rise to an O/XVS pattern. 27 Let us start with object fronting in Mío Cid (27) Batllori and Hernanz (2015), despite calling this type of fronting 'weak focus', they do not say that it is indeed a focal element. The term "weak focus", as explained by the authors, derives from the term "unmarked focus" that Benincà (2004) associates to a projection lower than FocusP that is neither a Contrastive Focus projection nor a Topic one. As such, it is perfectly compatible with the non-focal fronting terminology used by Leonetti (2017). Importantly, for Sitaridou (2011 et seq.) this type of fronting is identified as informational focus; crucially, it does not display any quantificational properties (essentially following Rizzi 1997; Kiss 1998, among many others) as is the case with contrastive focus; therefore, it is compatible both with Batllori and Hernanz' (2015) 'weak focus' and could also be envisaged to create one single partition with the verb as claimed to be the case by Leonetti (2017). However, for Sitaridou (2011), this position is claimed (i) to be immediately above the TP and (ii) to be associated with syntactic properties such as adjacency with the verb; thus, these two points diversify her analysis from the above-mentioned authors. 27 Note that OVS can also be interpreted as contrastive focus fronting in Old Spanish but we are not dealing with this type of focus here (but see Eide and Sitaridou 2014 As previously mentioned, according to Leonetti (2017: 907-908), non-focal fronting does not trigger any informational partition since there is no brake in the intonation of the utterance, and the consequent inversion of the subject does not yield any special interpretative (aka focused) effect on the inverted subject. In Leonetti's words (2017: 922), "fronting usually gave rise to sentences interpreted as single informational chunks, either with wide focus readings or with emphatic values derived from Verum focus". In a similar vein, Sitaridou (2011; and Batllori (2015) (among others) agree that the interpretation of the fronted constituent depends on the context -it may be focal, it may not, with more than one focal interpretations available ranging from contrastive to informational/broad/verum focus -hence differs from formal movement where the fronted constituent does not necessarily have a discourse flavour.
Crucially, object fronting has become much more restrictive in Modern Spanish (Sitaridou 2011;Leonetti 2017: 927 (27) supports, in our view, the independently argued assumption that Old Spanish could have had a preverbal information focus position (see Sitaridou 2011: 178;2015: 133;Batllori 2015; and even Elvira 2017 who argues in favour of the informational focus value of the colon). If so, in this respect, our analysis differs from Leonetti (2017) who, although he does not explicitly commit about syntactic positions, he seems to favour a single position for the fronted material. Leaving aside for a moment where Leonetti's single position might be (spec-CP or spec-TP?), in the previous sections we came to the conclusion that certain fronted expressions can be either in the left periphery or in spec-TP (as is still the case in Modern Spanish). Nevertheless, we are still left with cases of object fronting for which neither the spec-TP nor a quantificational focus position in the left periphery seems to be the right host. So, in Old Spanish either there is an additional focus position immediately above TP, or the spec-TP could have hosted fronted objects whereas, clearly, it can no longer do so in Modern Spanish. We find the latter less plausible.
Let us now consider adjunct fronting which could still be acceptable in narrative Modern Spanish without implying any IS partition value (28). In contrast to adjunct fronting which has been preserved, adverb fronting (29), which demonstrates undisputable cases of verum focus, has survived through the fossilization of these structures and subsequent reanalysis as emphatic polarity particles such as bien 'indeed', and ya 'indeed' in Modern Spanish (see Batllori and Hernanz 2009;2011;2013;Rodríguez-Molina 2014 With regard to the fronting of non-finite verbal forms (32-34), most modern equivalents would be ungrammatical in Modern Spanish. These structures have been extensively discussed (see Lema and Rivero 1989;1991;Rodríguez-Molina 2010;Sitaridou 2015;Batllori 1993;Elvira 2015; Octavio de Toledo 2015, among others) and all authors (with the exception of Lema and Rivero 1989;1991) agree that the IS value of non-finite verbal fronting in Old Spanish correlates to either contrastive, broad or verum focus depending on the context without necessarily sharing the same position (see Sitaridou 2015: 132, cf. ex. (48) and (50) Estos luego corriendo fueron a los grandes palacios these immediately run.prog go.prf.3pl to the big palaces a se lo decir to =dat.3sg =acc.n.3sg say.inf 'And these went hurriedly to the big palaces to tell him'

Diachronic implications and conclusion
In this paper, we have presented a cross-genre study on word order variation and discourse organisation in Old Spanish texts. A comparative analysis focused on fronting phenomena with the aim of evaluating their discourse role and assessing possible influence of textual genre so that we can ultimately establish what the changes have been diachronically -see Table 1 for a summary.
We have found out that: (a) stage topics and deictic fronting has been stable diachronically; however, it must be noted that in our texts -and above all, in epic poetry -the stage-topic configuration coexists with another one that involves a projection in the left periphery (see d, below); (b) quotative inversion has also remained stable in the history of Spanish largely due to the nature of the semantics of the verba dicendi and other verbs of communication with some variation again attributable to genre. As to be expected, in quotative inversion, OVS is not generally favoured in Old Spanish (or Modern Spanish for that matter) given that the object is usually clausal hence heavy and thus postverbal); instead this type of inversion mostly denotes narrow focus on the subject; (c) with regards to quantifier fronting, we have identified two types, one of which conveys verum focus; this type has been preserved until today; (d) the type of fronting which has changed most is without a doubt information/broad/weak focus fronting of both objects and non-finite verb forms thus supporting the idea that (e) Spanish has lost one focus position specifically designated for information/broad/weak focus in the left periphery. In the few cases it survives it has become more restrictive in Modern Spanish and it does not denote any broad focus anymore, but has instead grammaticalised further into verum focus (or into a synthetic tense in the case of the futures and conditionals).
On the basis of these findings, which shed light on the organisation of information structure, we can now start approaching the issue of diachronic change. In Sitaridou (2011;2012: 593-597) it was claimed that, diachronically, "although word-order patterns do not become obsolete, they, nonetheless, specialise in terms of what discourse reading they may convey and, in this sense, it can be claimed that word order becomes more fixed although, sensu stricto, it is not the word order which becomes more fixed, but rather how the discourse readings map onto syntactic positions." Our findings seem to confirm this. Moreover, Sitaridou (2012: 597) claimed that "in an OV grammar of the Latin type, both discourse roles of old and new information would be encoded, necessarily, to the left of the verb. However, in a grammar like that of Old Spanish, whilst optional movement makes OV (syntactically) opaque, the preverbal position is still related to both discourse readings of old and new information, as was the case in Latin. However, as VO was becoming fixed, the competition between the two distinct orders forced their discourse specialisation: the VO order conveyed informational focus, whereas the OV order became marked and had to be associated with the most marked reading, namely contrastive focus." Our findings in this paper also seem to offer further support to this claim. Specifically: (a) If Classical Latin, by virtue of being an OV language, had a caret feature namely purely formal, arbitrary diacritic without any semantic content which moved the object to the left of the verb (see Biberauer et al. 2014); and (b) If "it is the formal features of the target language that are not instantiated in the L1 or have a different setting, (which) cause learnability problems" (Tsimpli & Mastropavlou 2007: 215), it is possible that OV continued to be used in Old Spanish, but as a result of focus-driven movement for information focus thus rendering the prior caret feature of Latin into an interpretable one. Then, it is possible that the preverbal focus position was yet again reanalysed as for hosting only the most marked of the focused constituents, namely the ones conveying contrastive focus whilst informational focus could now be rendered postverbally with the aid of nuclear stress. Clearly, the diachronic account is to be developed further.
ances and Texts of the Royal Scriptorium Manuscripts of Alfonso X, el Sabio, ed. de Ll. A. Kasten y J. Nitti, Madison, Hispanic Seminary of Antiguo Studies, 1978 (microfichas)