DEFINITENESS STRATEGIES AND WORD ORDER IN EXISTENTIAL-LOCATIVES AND LOCATIVES IN LATE AND VULGAR LATIN

§ 2. The two parameters are still subject to investigation. The discussion of definiteness markers focuses on how and when precisely the article appeared, but it has yielded contradictory conclusions: while some scholars acknowledge the existence of articles in Late and Vulgar Latin texts, such as Egeria's Peregrinatio,4 others deny it. In addition, there are some less extreme views arguing for an intermediate stage in the long evolution from demonstratives to articles in Late Latin texts. 6 The research on word order, especially in Late Latin, frequently focuses on the position of the elements considered basic in the language; 7 thus, generally speaking, almost everyone is agreed on the SOV> SV0 change, although an altemation of the two orders can be established for the Classical as well as for the Late period ([S]OV /[S]V0).9


O. INTRODUCTION
§ 1.The parameters which began to undergo a profound change in Late Latin include the marking of definiteness and the gradual fixation of a different word order. 1 These two phenomena are brought into connection by M .. Durante's observation (1981, 62)  that article development 2 is one of the main agents involved 3 in the emergence of a fixed order of constituents.§ 2. The two parameters are still subject to investigation.The discussion of definiteness markers focuses on how and when precisely the article appeared, but it has yielded contradictory conclusions: while some scholars acknowledge the existence of articles in Late and Vulgar Latin texts, such as Egeria's Peregrinatio, 4 others deny it. 5In addition, there are some less extreme views arguing for an intermediate stage in the long evolution from demonstratives to articles in Late Latin texts. 6he research on word order, especially in Late Latin, frequently focuses on the position of the elements considered basic in the language; 7 thus, generally speaking, almost everyone is agreed on the SOV> SV0 8 change, although an altemation of the two orders can be established for the Classical as well as for the Late period ([S]OV /[S]V0). 9§ 3.With regard to Late and Vulgar Latin, however, there are only a few specialised papers dealing with potential historical changes in the position of the verbs which present special difficulties in the language, not only with their complex semantic content but also with their ability to form multiple constructions. 10This is the case with the verb sum in structures expressing a semantic notion of location.The relationship between word order and definiteness in these structures has been exarnined by scholars attempt-ing to explain the difference between the so-called 'locative' constructions (Loc) and 'existential-locatives' (ExL), 11 both containing the verb sum: while the former typically front the location-encoding term (p) with a definite subject, the latter display the opposite distinctive features (a S[-Def] 12 appearing in the pS sequence).13l.THE PURPOSE, CORPUS AND STRUCTURE OF THE PAPER § 4. The present paper focuses on the two above-mentioned parameters, which are of interest because of the possible interrelation in their development.The issues outlined above can be studied particularly well on the locative and existential-locative structures.§ 5.The research corpus consists of narrative and/or historical texts (Peregr. 14and Oros.hist.1-5).The reasons are: (i) these are subject to fewer metrical and stylistic restrictions than other types of text, (ii) the content itself demands the inclusion of II Cf.E. v. CLARK (1978: 91-101); T. GIVON (1978: 306 ff.); s.KUNO (1971); C. LYONS (1999: 88-89); CA-BRILLANA (2001).In Classical Latin, where the predictions ofword order are not fulfilled, these structures may be exemplified by the following: Loc: ubi nuncficus Ruminalis est (Liv.1.4.5);ExL: vastae tum in his locis solitudines erant (Liv.1.4.6). 12However, the princip le ofthe subject being ±Definite does not work unequivocally and universally.Since the analysis seems to confirm the definiteness restriction (cf.e.g. C. LYONS (1999: 236-246), who implies a certain correlation between the (in)definiteness ofthe subject and the typological distinction established between the locative and existential-locative construction; indeed, the existential constructions oflanguages which possess special markers for the ExL structure (there) are presented as granunatical contexts which confirm the indefiniteness of a term), the rare examples displaying the inverse correlation (Loc: TS [-Defj :: ExL: TS [+Defj) are ofparticular interest.The presence of a [-Defj subject in locative constructions is a well-documented interlinguistic phenomenon; this fact is quite bizarre in terms of regularity, but in no way seems to threaten any restriction of granunaticality ( cf. C. LYONS, 1999: 23 7); thus: ultimum orationis fait, se arma capere, vocare omnes Quirites ad arma; si qui impediat iam ". se Publio Valerio consularis imperii, iam tribuniciae potestatis sacratarumque legum oblitum, quisquis ille sit, ubicumque sit in Capitolio.in foro.pro hoste habiturum (Liv.3 .17. 7).This example reveals the anomaly of a locative expression whose subject possesses a low degree of definiteness due to its non-specific, indefinite reference (qui impediat, ... quisquis ille sit), but the typological parallel clearly suggests a locative interpretation ('any who resists [se.taking arms ], whoever and wherever he may be, on the Capitol or in the Forum, ... ') rather than existential-locative ('*there exist some people who resist wherever they are, on the Capitol or in the Forum, ... ').By contrast, the presence ofa [+Defj subject in an existential-locative structure threatens the definiteness restriction not only as an anomaly but asa 'non-grammaticality' in some languages ( even in languages Jacking definite articles: cf. the argument ofHuANG (1987) ap.C. LYONS (1998: 240) on Mandarin).However, C. LYONS (1999: 239) points out that the definiteness restriction is far from being as categorical in English as it is usually assumed in linguistic discussions, and its expression is particularly complex in Latin (erat tum inter equites tribunus militumA.Cornelius Cossus ... (Liv.4.19.1)(?)).As the extensive research into existentials has underlined, the presence ofa [+Def] subject in an existential-locative structure is made possible by exceptional communicative contexts (' listing-', 'reminders-', 'amount readings': cf.C. LYONS  (1998: 241-246)) which assign, in terms ofFunctional Grammar, the pragmatic function ofFocus to the subject.The problem raised is stili an open question.Cf.M. DiAz DE CERIO (2002). 13This is what could be called the 'Hypothesis ofDefiniteness'.On the behaviour ofthis Hypothesis in Classical Latin andAncient Greek, cf. C. CABRILLANA-M.DiAz DE CERIO (2000).Ofthe Vulgar Latin texts, the prediction tends to be fulfilled in Egeria (Loc = Sp: 61.53% :: ExL = pS: 61.12%) but not in Orosius, where the most usual sequence is pS regardless ofthe kind ofconstruction (Loc: 71.43%; ExL: 63.64%). 14The analysis includes only the first 23 chapters-that is, the narrative partas the most exhaustive source oflocative constructions, since the part describing liturgical rites yields far fewer examples.As corroborated by R. LAPESA (1961: 26) or A. NocENTINI (1990: 149), the liturgical part contains fewer demonstratives as well, for the writer does not proceed in the same way as when relating personal experiences.
topographical digressions ( excursus ), 15 which invite the use of otherwise less frequent 16 locative structures, and (iii) the period of the selected texts falls within the larger frame17 proposed as the most significant in the evolution ofthe key parameters discussed in this paper.At another level, the texts -the Peregr. in particular -show an unavoidable, more or less conscious overlap of functions and tendencies typical of a transition period,18 as well as a mixture of grammatical levels or systems. 19As a foil for evolutionary comparison, 1 have selected a text from the Classical period (Liv.1-4).§ 6.The paper starts with a brief explanation ofthe theoretical concepts used in this research (2) and considered suitable to carry out the analysis (3).A general presentation of the data gleaned from both texts (3.1) is followed by an examination of the different methods indicating the definiteness of a term in Latin (3.2), with a special focus on demonstratives (3.2.2).The final section (3.3) tackles the order ofthe basic constituents in locative constructions, in order to establish the relationship between the two analysed parameters.
2. PRELIMINARY CONCEPTS § 7. The inclusion of the definiteness (Dej) analysis calls for a clarification of the concepts assigned to that parameter, a task which presents difficulties both in its own right and in the context of the proposed research.The aim of this paper, however, is not to provide a theoretical approach to the above-mentioned parameter, but rather to examine its influence and behaviour in the texts.The discussion will therefore limit itself to clarifying the sense in which some of the concepts are used.§ 8. Definiteness is necessarily related to variables belonging to different linguistic levels: (i) to the semantic level, through the property of referentiality; (ii) to the grammatical level, through the formal reflection of determination.Differing from language to language, the system of determination is apparently responsible for the listener's identification of an entity as ±Definite.
The referential terms that will be considered are those able to process ±Def, that is to say, those relating linguistic expressions 20 to real or possible entities (referents). 21Among them, we will assume that indefinite terms ([-Def]) refer to entities which are not identi:fiable by the listener, in contrast to definite terms ([+Det]), which refer to identifiable entities and enable a 'referent identification'.
Conversely, the identifiability of a referent by the listener has been approached from the pragmatic point ofview.According to this, various 'sources of availability' 22 can be established from which the listener obtains information: (i) long-term pragmatic information/general knowledge ('the sun', 'Paris'); (ii) current pragmatic information introduced in the preceding discourse ('I have bought a car; you will see it tomorrow'); (iii) information which is perceptually available in the situation ('do you see the man with the green sweater?'); and (iv) inference from the information available in any ofthe sources (i)-(iii) presented above ('a submarine ... the hatch').§ 9.Among the grammatical reflections of definiteness, the presence/absence of the article is the most prominent, but not the only one.Other marks of definiteness are formed by lexical and word-class distinctions.Of these, lexical differentiation 23 exceeds the boundaries of the present study, but as for word-classes, it is proper nouns, demonstratives, 24 etc. that are particularly operative.§ 10.The relationship between the pragmatic and grammatical levels originates from causality: what causes a term to display some of the above-mentioned distinctions, making it ±Definite, is the availability of its referent to the listener, from one or more of the four basic sources.For example, what makes a proper noun or noun with only one referent assign ±Definiteness to a term is general knowledge.On the other hand, the ability to identify a referent from the previous context is demonstrated by the use of personal pronouns, demonstratives, relatives, and the like.
In order to carry out the proposed survey, we must therefore examine what impact the ±Definiteness of the terms involved has on the discourse.In this, then, we count on extemal <lata to establish the degree of development attained by the grammatical markers of definiteness in such texts.§ 11.5 This ratio can be explained by the following factors: (i) The type of passage examined, which is more descriptive throughout Egeria's text than in the five books of Orosius, except for hist.1-2.
(ii) The overuse of the passive voice, with verbs that define more fully the simple value oflocation expressed by the esse forms. 26iii) The consequent increasing use of the verb esse as an auxiliary form in the passive,27 resulting in a gradual loss ofits wide semantic range and restriction to copulative-identifier structures.
(iv) The author's characteristic conciseness, fairly often leading to the suppression of the verb: not only in copulative-identi:fier constructions, which are transformed into the so-called 'pure nominal'28 structures, but also in constructions where the verb form would normally operate with a location value. 29v) The use of alternative structures, typical of this language period, to convey notions including location.The above data yield the following hierarchy of definiteness-encoding procedures for the three corpora: Since this discrepancy can be partly explained with Orosius' tendency to imitate classicism, it seems suitable to analyse each non-classical text separately.The data shown above indicate the frequency ofthe procedures which, to a larger or lesser extent, establish a degree of definiteness in the subject term.Before any further discussion, however, it is necessary to clarify some points, particularly with regard to the procedures most frequently employed:

Markers of ±Def
1. Contrary to Late Latin texts, Classical prose is characterised by the absence of grammatical definiteness markers.This implies an expectation that, despite the lack of lexical or grammatical markers, the reader/listener will be able to identify an entity as definite because of the preceding context.
2. On the other hand, the use of proper nouns -either as nuclei or as determinatives in the genitive-is revealed as one of the most e:ffi.caciousand unmistakable markers of definiteness.This procedure does not exclude others but can in fact combine with them; thus, for instance, this marker is in many cases coupled with that of general knowledge (1) or of the preceding context (2): (1) (iunctus est cum eo loco) quo sunt Memoriae concupiscentiae (Peregr.1.1): "(next to the spot where the mountains open out is the place ofthe 'Graves ofCraving"' (2) (in eo loco) ubi fuit domus sancti Abrahae (Peregr.20.3): "first he took us to a church outside the city;" ( ... ) (ecclesia) ubit fuit primitus domus Abrahae (Peregr.

20.5):
3. The use ofthe relative pronoun as S opens two possibilities: (i) the definite quality ofthe antecedent (if present) spreads to the relative pronoun (3); (ii) the definiteness of the S is the result of the antecedent being implicit in the relative pronoun of a nominal relative clause.In the latter case, it should be pointed out that, while relatives with no antecedent can appear in Livy -( 4) -, the texts by Egeria and Orosius always contain an antecedent with a degree of definiteness which is completed with the relative clause [( 5)-(5')].In the ranking shown in Table 3, this procedure is listed in the third place, but its frequency varies: it amounts to 20% in Livy, while Orosius and the Peregr.display the similar figures of 9 .10%and 9 .80%respectively.

Demonstratives § 17.
The frequent use of demonstratives as Def markers in Egeria's text is by itself an indication ofthe gradual loss oftheir deictic function.This development may be clearly perceived in the cases of ille and ipse.According to Nocentini, the two demonstratives tend to become associated with known information, which may or may not have appeared in the preceding discourse.Thus they develop into increasingly emphatic signals for the listener to identify the entity involved, as seen in the following example: (8) requisivi ubi esset puteus ille ubi sanctus Iacob potasset pecora, quae ... Et ait mihi episcopus 'in sexto miliario est hinc locus ipse iuxta vicum, qui fuit ... ' (Peregr.20.11)In the case of ( 8)-an answer in direct speech-, the information expressing the subject is underlined by the use of ipse with locus and thus rendered as a definite entity, as is appropriate for the subject of a Loc construction.This example confirms the scheme proposed by Nocentini (1990: 146), who marks the features of ipse as [+detto +testo] in contrast to the [-detto +testo} of ille.However, there are also examples which seem to contradict this analysis: (9) requisivi a sancto episcopo, ubinam esset locus ille Chaldeorum, ubi ... Tune ait mihi ipse sanctus episcopus: 'locus ille, ... , est hinc intus in Persida .. .' (Peregr.20.12)This 'contradiction' actually reflects a fluctuating use, which can be enhanced by the co-existence ofmore than one 'system' in the Peregr., as stated above. 34§ 18.The use of demonstratives assigns the property ±Def also to terms expressing location, especially where the +Def of such terms contrasts with the indefinite quality characteristic of the subject in ExL constructions.In (1 O), the first occurrence of a yet unknown entity -'a hill' -in the text contains no (in)definiteness markers, whereas its second appearance, as of information already known, carries the determinative ipse.In the latter structure, the new information is conveyed by the subject, which is accompanied by a marker of genericity (plurima): (1 O) et quoniam inde ad sanctam Teclam, qui locus est ultra in colle sed plano, habebat .... ( ... ).Sed ut redeam ad rem, monasteria ergo plurima sunt ibi per ipsum collem (Peregr.23.2-4). 35xamples of these and other demonstratives displaying the contrast between unknown and known information are ( 11)-( 14): (11) accessimus ... ad eum locum, ubi steterat sanctus Aaron cum septuaginta senioribus, ....In eo ergo loco. 36 It seems clear, then, that there is a tendency-especially in the Peregr.-to associate a known and definite entity with the presence of certain demonstratives whose deictic function is weakening and shifting towards a 'phoric' one, thus forming a prelude to the Romance article. 37The indefinite and unknown entities, by contrast, usually display an absence of markers. 38Sometimes, however, they are marked by an element that seems to havelost its quantifying value to acquire another, which will in tirne become characteristic of the indefinite article: 34 Cf.§ 5. 35 Although the strategy is generalised particularly by Egeria, it is not exclusive to her: mons Caucasus inter Colchos, ... , et inter Albanos, ... , primum attollitur.( ... ).Itaque ipse Caucasus inter Colchos et Albanos, ... (Oros.hist.1.2.36-39).The contrast between the initial absence of markers and the subsequent addition of ipse can also be observed in locative structures with verbs other than sum: haec est auten vallis, in qua factus est vitulus, qui locus usque in hodie ostenditur: nam lapis grandis ibi jixus stat in ipso loco (Peregr.2.2). 36 The preceding context reads: accessimus ... ad eum locum, ubi steterat sanctus Aaron cum septuaginta senioribus.On the particulars ofthe prepositional phrase in eo loco, sometimes followed by a specification (in eo loco ubi fuit domus sancti Abrahae, Peregr.20.3), and its possible adverbial nature, cf.E. LOFSTEDT (1911: 143); R. AMBROSINI (1955: 101); V. VAANAN'EN (1987: 25-26).On the occurrence ofthe [N+demonstrative+relative] pattern in the role ofan article, cf.H. RosEN (1994: 131). 37From this perspective, H. RosEN (1994: 145) states: "The date ofthe emergence ofthe definite article in Latin may be pushed backward, much earlier than the communis opinio has it, to the fourth century." 38Cf. § 8.

57
(15) nune autem ibi nichil aliud est nisi tantum unus lapis ingens Thebeus (Peregr.8.2)39 3.3.Constituent order § 19.A discussion of this parameter must consider also the constructions where one of the constituents has a fixed position: cases where the subject (S) or a locative expression (p) is encoded by a relative pronoun or adverb.Given the generally accepted evolution SOV> SVO, the subject oftransitive constructions would not be expected to alter its position.On the other hand, the present study examines intransitive constructions as well.Thus the analysis basically focuses on the position of the verb, absolute and relative (that is, relative to the S and p ).This approach allows us to ascertain whether intransitive verbs, particularly exceptional ones like sum, undergo a similar historical change of position as transitive ones.§ 20.After excluding the constructions whose subjects are bound to the initial position, we obtain the following data on the absolute and relative positions of the verb:  °""'"' llllVS llll!SV § 21.The data shown in the tables above provide the basis for several conclusions: 1.The tendencies observed are clearer in Loc constructions than in ExL structures.2. There is a shift of the verb towards the medial position, which is much more pronounced in Egeria than in Orosius.The discrepancy is probably due to the latter having been more strongly influenced by the literary conventions of the Classical period, which established a fixed position for certain constituents.
3. Sirnilarly, there is a growing tendency towards replacing the sequence SV with the inverse VS. 40 It seems to be more common with sum than with other verb types. 412.In connection with the last point, 1 believe that we must focus on the smaller proportional difference between the two sequences in the Peregr., which displays a seemingly unmotivated altemation of the pattems Sp V /SVp and pSV /p VS.An exarnination of the individual examples of these sequences in Loc and ExL constructions reveals that the relative position of their constituents is, in fact, 'to a certain extent' indifferent, because the 'speaker' enables the 'listener' to identify the located entity (the subject) in Loc constructions as definite by (i) the assignment of a de:finite demonstrative, or (ii) the use of word-classes with a high degree of definiteness per se, such as proper nouns.The subject ofExL constructions, by contrast, lacks all markers of definiteness: It may be concluded that the presence of definiteness markers allows a certain freedom in the order of constituents.Sometimes, however, the listener's extra-textual knowledge of an entity, as well as its previous occurrence in the text, enables it to be identi:fied as definite on its subsequent occurrences.In such cases the presence of a grammatical marker is not necessary, and the order of the terms whose position is not fixed (Sand V) may vary: (19a) Moyses ... locutus est ei Deus de robo in igne (Peregr.2.2) (19b) denuo ad illud caput vallis descenderemus, id est ubi rubus erat (Peregr.
4. CONCLUSIONS § 25.From a global perspective, the <lata in this corpus allow us to perceive an evolution in the order of constituents employed in the examined structures -one that is more evident in Egeria than in Orosius.There are tendencies to place the verb in the medial position, to make increasing use ofthe VS sequence at the expense ofthe Classical SV, and to include the verb sum proportionally more often than other types ofverb.§ 26.Among the different procedures for encoding definiteness which are not mutually exclusive, the use of demonstratives, scarcely present in Livy, gains ground and reaches the highest frequency with Egeria.As evidenced by examples ( cf. § 2.2.2), demonstratives -especially ille and ipse -gradually lose their prominent deictic function, a process which is still unfinished in the period examined here.This mutation is associated with factors of a pragmatic-contextual nature, which assume a key role in the use of definiteness markers, and thus in the distinctions between the analysed structures in Late Latin.§ 27.Undoubtedly, the tendency to associate definiteness with the presence oflexical markers developed in Late and Vulgar Latin and vice versa seems justi:fied.It is, however, possible to reason as follows: 1.The word order ofClassical Latin is generically defined as free but not indifferent. 46At this stage of the language, with no complete lexical or grammatical system to mark definiteness, it runs counter to what is predicted by the Hypothesis of Definiteness: 47 it does not play a truly distinctive role in Loc constructions, at least not in a clear and universal way.Nevertheless, the 'listener' is able to identify the ±Definite entities.
2. During the transition from Late Latin to Romance languages, this word order gradually became fixed due to the progressive loss of declensions; moreover, a lexical and grammatical system evolved to mark the ±Definiteness of terms.This system, however, could not replace word order in a role which the latter did not play.

Povzetek
DOLOČNOSTNE STRATEGIJE V BESEDNEM REDU IN V EKSISTENCIALNO-LOKATIVNIH TER LOKATIVNIH ZVEZAH V POZNI LATINŠČINI Cf.s. C. DIK(1989: 114).23As in Classical Latin, lexical differentiation is practically non-existent in the analysed texts, except for the Preliminary data § 12.The sum total of the locative constructions is distributed between the two texts as follows: Ev en a cursory glance reveals that the to tal number of examples is signi:ficantly higher in Peregr.than in hist.
Quantifier]32 III.lnference § 15.The subject term analysis ofthe corpus selected for this research, compared to a text from the Classical period, reveals the following distribution:

Table 2 :
Strategies for Encoding Definiteness of the S Strategiesfor D~finiteness

Table 3 : A Quantitative Hierarchy of Strategies for Encoding the Definiteness of the S Livy
App./Posses.Pro./lnfer.§ 16.Although some examples display more than one source of availability, 33 the Peregrinatio is clearly dominated by demonstratives, in contrast to Livy and Orosius.