Volume 9, Issue 2

Though the Latin language seems to have been rather stable and have undergone few changes even in the passage from early to classical Latin, some new words and morpho-syntactic forms entered into Latin, some perished and some other were reintroduced. Two forces, therefore, seem to have been active in Latin which can be calied enrichment and simplification (enrVsimpl.). After taking into account this Situation of Latin language I will use this criterion to explain some syntactic enlargements, and in particular the birth and enlargement of optative and subjunctive moods and the passage from an early phase of Indo-European languages (as represented by Hittite) to a historical one (I will give a hypothesis of such a birth). Birth and expansion of these moods offered the possibility of distinguishing deictic and non-deictic verbal processes. Such an enrichment of Greek and Latin fmite verb produced some distinctions which could not be expressed before and perhaps didn't exisL This brought to a syntax more and more concentrated on the use of finite verbal forms and permitted an expression which got more clear and fine and, through conservation of old expressions and introduction of new ones, was also richer.

This paper is a refutation of the existence, in Latin, in addition to the traditional "active" and "passive", of a "middle voice" (expressed by analytic means e. g. se movet), which has been put forward recently. On the contrary, it will be shown that the constniction "active verb + reflexive pronoun" is but an ordinary use of the active (transitive) voice, in which an anaphoric counteipart of the subject (i.e. se) has the function Objec. Consequently, a sentence such as se movet brings about an "objective", resultative view of the event, as any active (transitive) sentence does, in Opposition to the "subjective" and "existential" representation of it which is expressed by a "passive" form like movetur. The demonstration will be supported by an accurate Observation of contexts in which the "reflexive construction" and the "passive" form of a number of verbs are employed respectively.

Settling a clear distinction between pronouns (viz. relative or interrogative pronouns) and conjunctions is not very easy in many languages, peculiarly in Latin. Pronouns, conjunctions and their correlatives are derived from the same kinds of stems and they have similar uses in clauses. We propose to discriminate what we call an endophoric relation for conjunctions and what we call an exophoric relation for subordinating pronouns.

Few studies have dealt with Julius Valerius' language. The aim of the paper is to analyse the use of the nairative Infinitive and of the historical present in the Res Gestae Alexandri Macedonis. Three aspects will be considered: their respective distribution in the Romance, their "in series" occunrences and the syntactical constraints that characterise the narrative infinitive. The Observation confimns what has already been noticed about both of them.

Functional Verb Constructions have found increasing interest in Latin linguistics since the valency studies at the end of the seventies. In particular the adferre + acc. type of an abstract verb noun is of great interest in its causative function. Beginning with a precise definition to eliminate superficially similar phenomena, e.g. manus adferre (1.). part 2 of this study is a synchronic analysis within ciassical prose. An overview illustrates the high proportion of FVCs in Cicero's coipus (2.1). By examining the two most frequent examples, the different syntactic functions of the transitive vim adferre (2.2) and causative spem adferre (2.3) become apparent, whereby the complex structure of the causative type is revealed through valency analysis. The question as to the conditions under which causativity is blocked (2.4), finds its answer in transitivity and the thematic role of the dative (2.4.1). Nevertheless, alternative causative constructions in Latin, e.g. facere + Infinitive, seem to have similar restrictions (2.4.2). An overview of the usage before and after Cicero in part 3 shows that this FVC can be found in all central periods of Latin, but especially in authors of elevated prose.

In this contribution the linguistic behaviour of iam in negated utterances is investigated on the basis of the analysis of iam developed by Kroon & Risselada (1998; 2002). It is argued that the focal and polar uses of iam as distinguished by Kreon & Risselada cannot interfere with negations: iam behaves in exactiy the same way as in positive utterances. The temporal and scalar use of iam, however, both characterized by the feature of phasality, are in positive as well as in negative contexts part of the phasal polarity system. In these cases, iam in negated utterances carries the meaning 'no longer' and as such it is described in a contrastive analysis with non amplius. However, non iam may as well mean 'not yet'. As such it is set against nondum and adhuc non. It is argued that non iam in this meaning is very infrequent and confined to a special pragmatic use in which the feature of counterpresuppositionality is veiy much to the fore.

The aim of the paper is to describe types of anaphoric relations that hold between sentences in Latin, and to discover diachronic shifts that could happen in the use of these types. Anaphora can be realized by pronominal chains or by lexical repetitions; some connecting elements, especially conjunctions of coordination also function anaphorically in the text Often this type of binding is optional, and it insists on spatial and temporal frames of the events related. Lexical means become more frequent in Late Latin, and optional connection appears also more often. These phenomena can be evaluated as stylistic deviations, but it is probable that they are related to deep syntactic modification of postclassical Latin. Documentation is taken from classical and post-classical authors and from the Bible.

Descriptions of the Latin particle quidem tend to centre on the concept of adversativity or contrast. In this paper I propose an alternative account in which quidem is described not as a marker of a specific type of semantic relationship, but rather as a fairly neutral marker of the discourse Organization. More specifically, it indicates that its host unit is to be regarded as a constitutive part of a larger conceptual whole. This account is based mainly on the Observation that in a heterogeneous corpus of about 1000 instances quidem occurs significantly often where grammatical structure and discourse structure do not coincide, that is, in those environments in which communicative autonomy clashes with semantico-syntactic dependency.

Zusammenfassung "Nominalisierter Relativsau" (NR) heißt hier der Relativsatz, der auf kein lexikalisiertes Normen bezogen ist, eine Gesamtheit von Wesen (qui) oder Sachen (quod) bezeichnet, im Satz die syntaktische Funktion und den Wert eines Nomens einnimmt. Hierbei kommt das kataphorische Pronomen is, ea, id oft (aber nicht immer frei oder zulässig) vor, namentlich (und nicht nur) um die syntaktische Funktion des NRs ins Klare zu bringen. NR verteilen sich auf zwei Haupttypen: Referenz-orientierte NR, einerseits, die eine Antwort auf die Frage Qui homo / Quae res! geben, Wesen oder Sachen bezeichnen, das kataphorische Pronomen vor- oder nachgestellt immer zulassen, die Modalopposition einhalten, wobei dem Indikativ die Fähigkeit zukommt, das Referent des NRs nach der Frage Quis / Quid! zu identifizieren, dem Konjunktiv die Anlage gehört, Virtualität des Verbalprozesses, Indetermination des Referenzträgers anzudeuten; Generisch-orientierte NR, andererseits, die auf die Frage Cuiusmodi homo / res? antworten, Klassen von Wesen oder Sachen beschreiben, Gebrauch des Indikativs, kataphorisches Pronomen im Nominativ und nicht-präpositionnellen Akkusativ immer ausschließen. Ergänzung mit omnis gehört zu der Referenzorientierung des NRs.

Classical historians use the structure of type ut posset resisti. The syntactic analysis of the passive infinitive isn't obvious : is it a predicate (of an infinitive clause) or isn't it ? The functional reflexion below proposes arguments pro and contra these two analyses.

Several prepositions can denote Cause in Latin. They all have a basic spatial meaning from which the causal meaning is derived through different metaphors. Their usage and distribution varies among authors and through time. Related semantic roles, such as Beneficiary and Purpose, can be expressed to varying extents by the same prepositions that express Cause. In Late Latin Cause and Beneficiary (and later Purpose) merged completely, a development which is still partly visible in the Romance languages. It is argued that merger of these semantic roles is favored by the fact that the prepositions involved expressed some sort of static Location as their original local meaning, and did not denote directionality.

Modem languages may generate possible worlds by means of conditional mood: in what follows we analyse how the Latin language can obviate the lack of this mood. We shall consider the not-actualisation by epistemic modifiers which operate on factuality, as the potential. When the Speaker wants to point out the lowest degree of actualisation, the perfect subjunctive is used to kindly mitigate an assertion or to emphasize “le caractère invraisemblable ou excessif” (H. Vairel 1980: 130). We shall analyse also the "untrue conditionals" which introduce not-conditioned probable worlds and the idea of not-actualisation, rejecting the predication into the "anti-universe of the Speaker^ (R. Martin 1983).

There is a subgroup of the uerba accedendi that has the semes [+ etiam] and [+ dicere]. The unipersonal accedit, however, has the semes [+ etiam] and (+ euenit], As regards the lexemes that have the seme (+ dicere], they function as the uerba declarandi. But the verb accedit functions as a uerbum accidendi because of the seme (-)- euenit], The quod-c{dMS&, which depends on these verbs, bears factivity, white the «/-clause bears important or rhematic information. Conceming the veib accedit, the ur-clause generally has a real modality and it is commmutable with the quod-c\a.asc in Classical Latin, i.e. they can have the same propositional content. But since the l" Century AD the complementizer quod takes the place of ut, because this is in a factive frame constituted by the matrix verb, by the propositional content of the complement construction and often by a pronominal cataphora, also with factive value. It announces the complement and does not commute with the adverb ita, because this correlates with the moipheme ut, but not with quod. This displacement of the complementizer ut indicates that quod began to be established as a general subordinator.

On the basis of a selection of Latin-written Asturian documents from the Xlllth Century, we analyse the relation between subjunctive forms amavero, amavissem and to a lesser extent amarem in order to clarify the alleged neutralization of the corresponding Romance forms amare and amase as displayed in contemporary texts in vemacular language. Only two examples - deeply Romanised - can be found in the analysed corpus in which the content of amavero seems to have been asumed by amavissem, a fact which is undoubtedly related to (i) the improvement of late-medieval Latin with respect to that of previous centuries and (ii) its departure from spoken and written Romance.

Sununary Relative clauses in Juridical Latin are strictly connected with the succession of subjunctive and indicative mood and with the use of demonstrative and determinative pronouns. Regarding the first point, we observed that the Opposition between indicative and subjunctive mood often expresses the Opposition between factual and non-factual utterances. Subjunctive mood is also more frequent in the second part of the responsum where the jurist expresses his own opinion (so he does not assume his communicative responsibility), whereas the exposition of the particular case often presents relative clauses with indicative mood. In the same way our jurists (Alfenus Varus and lavolenus), purposely use at a very high rate the determinative pronoun is, ea, id, that has not a deictic force. In fact they often have to switch, in their responsa, from the specific case (that it is to say from the particular case) to the general one, to the universal case. On the contrary deictic pronouns, that occur at a very low percentage, are always used to localize something in a specific place and time and they are usually employed during the description of a particular Situation.

This paper deals with the use of Latin particles in a specific environment, viz. interrogatives (or questions in a Wide sense). For that purpose, I first explore the various functions which questions can have, both as individual speech acts and within larger discourse structures, on the basis of modern insights gained within conversation analysis and discourse analysis. It tums out that in interrogative utterances, particles are more often connected with specific discourse functions than with illocutionary functions, and they seem to be used more in particular to mark the various types of reactive questions that are found in the comedies of Plautus.

This paper investigates in which positions satellites can occur in the Latin clause. The theory of Functional Grammar (FG) assumes that four types of satellites are to be distinguished and that these types differ, amongst other things, in their formal behavior. In the following, predictions about the position of satellites in the clause will be checked with regard to a corpus of letters by Cicero (Ad Atticum, book I-II). The relative position of satellites, arguments and the verb will be examined with particular interest in the position of satellite clauses. It will be shown that predicate satellite clauses prefer to be placed ad the end of the sentence, while illocutionaiy and prepositional satellite clauses tend to occur at the beginning of the sentence. The type of predication satellites should be split into two groups: purpose, consequence and causative clauses predominantly occur after and temporal, conditional and concessive clauses before the main verb. These observations are discussed in view of some general principles of Functional Grammar.

The various levels of description to be discussed in the present paper are related to the socially relevant contexts depicted in our literary corpus. The scope of this matter is knowledge of use, which does not seem to be restricted by etymological relationships. From a glance at the quotations written, we are already familiar with the connection between deliberation and decision, despite their own characteristics. The existence of a large number of terms suggests that words rather than phrases are the units of sense. Nevertheless, the syntactic frame is no longer a mere option: it has become its structuring principle. Particularly relevant to it would be the relation assertive/report, that breaks up into two kinds of decisive (i. e. determination to act) expressions. Some of them are related to dare iudicium, but most can be seen as linguistic devices, enabling discourse organization. It should be stressed that these approaches appear to complement each other fruitfully, thus providing a better- defined (and linguistically based) description.

This article focuses on the distribution of the Latin reflexive personal pronoun in object-control constructions, also known as A+I. and in infinitival Acl-clauses. In object-control constructions, the accusative anaphor se never takes the position of accusative subject of an active VP inf . It is only found in the position of accusative object of that VP inf. Here two antecedents are possible - the accusative subject of the VP inf and the nominative subject of the matrix verb. The first phenomenon is known as local reflexivity, and the second phenomenon is described as middle-distance reflexivity.

The paper sets out the question of possible links between reports of speech acts and verbal nouns in Latin. The syntax of the verbal noun has mainly been investigated with respect to its governing force. I introduce the notion of the report of a speech act in sections 1, 2. A survey of earlier work on verbal nouns as complements is presented in sections 3, 4. I try to sharpen the focus on a type of complement clause involving the report of a speech act, in contrast to modes of indirect discourse usually considered in discussions on complement clauses.

Siunmary The V(erb)-S(ubject) order in Latin is traditionally considered as 'deviant' or 'abnormal'. However, VS ordering is due to various pragmatic functions. An analysis from a functional point of view shows that we can speak about 'emphatic' verb only if it carries the function of focus (contrastive, verumfocus...). Moreover, a sample of texts from the classical period shows that the verb occupies normally the initial position in the clause when the sentence serves to introduce a Future Topic, a new entity that will be further developed in the subsequent sentence. Introducing a Future Topic concerns especially existential and locative sentences containing the verb esse and sentences with verbs 'appear', 'remain' and similar. The pragmatic function of Future Topic entails normally the first position of the verb in the case of existential and locative sentences; the first position is only frequent for verbs like 'appear' and 'remain' which can figure also in the final position.

Latin grammars show a varied repertoir of the so called "impersonal" constructions. However under that label we find rather different phenomena exhibiting several regularity degrees. Together with real impersonal constructions, i.e., passive voice third person Singular forms, lacking subject, object and agent, other items such as weather and ethical verbs are added to the paradigm, though these conform a particular closed group that cannot be mirrored by others. Subjectless constructions formed by a third person plural active voice dicendi verb are also seen as impersonal; they are impersonal inasmuch as they lack a real subject, but they cannot be compared to real impersonal constructions, because only a limited number of verbs can be used in such a way: dicunt, ferunt, tradunt... They are not properly impersonal, as they have an indeterminate subject Finally as a special case of third person Singular active voice constructions some grammars mention certain sentences where we cannot find or elicit a subject. For these cases found in special typology contexts, I propose the name "apersonal constructions", as they show a personal formation, though no particular person is identifiable. We can provide a formal explanation for some of those cases, but not for others.

To whom the propositional content and its expression must be attributed in an indirect speech? My aim in this study is to examine different utterance strategies carried out by the narrator within a syntactically indirect speech: the narrator may partly re-delegate the deixis point of view to the characterutterer, or the propositional content might be interpreted as being valid for both speakers; on the contrary, the narrator might intentionally reiterate utterance features attributing the speech to a characterspeaker, even through syntactical discrepancies. All these positions will be examined in the framework of a wider description of the indirect speech.

In this paper I analyse the conditions for the use of repente / subito as discourse disjuncts. Attending to linguistic parameters such as type of text, tenses, clause type, collocations with particles, occurrences in parenthetical sentences and in clauses with VO word order, it can be claimed that these disjuncts are a way for the speaker-narrator to present new and unexpected events as prominent information in text units of central information.

We call correlation two different syntactic structures: First, when a subordinate clause is followed by a main clause beginning with a grammatical constituent of the same meaning as the subordinating conjunction. Then when a syntactic structure consists of two parallel clauses none of which seems to contain a subordinating morpheme, but both of which begin with a grammatical constituent of the same meaning. The first structure is a particular case of Subordination, the second is a particular case of juxtaposition, i. e. of coordination without a coordinating conjunction. Nevertheless both structures share a syntagmatic feature and an informative one.

Despite privileged linguistic relationships between proper noun and Singular mode, the former can have an inflected form in the plural. Following a referential approach, and reviewing a modern analysis, we establish a typology of the various onomastic plurals, and we study the linguistic Operation and conditioning suited to each different type.

The aim of this study is to investigate argumentative elements in the narrationes of Cicero's forensic speeches. Narrative texts are generally characterized by their thematic development, in which continuity of characters, time and place play an important role. In addition, the narrationes of forensic speeches are expected to be influenced by the overall argumentative text type (see Kroon 2000 for the embedding of text types). A prototypical argumentative stnicture is based on causal relations in general and particularly on conclusions. An extended analysis of the coherence patterns in three narrationes of Cicero (Pro Archia, Pro Milone and Pro Rege Deiotaro) reveals that a certain subtype of causal relations dominates this narrative subtype. The theoretical notions derive mostly from the work on coherence relations of Sanders, Pander Maat & Spooren (1992), who benefited significantly from the ideas on relational coherence of Halliday & Hasan (1976), Mann & Thompson (1986) and Sweetser (1990). A relevant aspect that is often ignored, however, in these publications is point of View on the causal relation. By including this element in my analysis, the results give more insight into the diversity between the narrationes, which can be explained by the dissimilar historical settings of the speeches. Another advantage of a combined causality/perspective analysis is an improvement in describing frequent causality markers (itaque, igitur, ergo).

By prepositions people can more precisely describe complex relations of persons and things than by means of pure cases. Precision was especially necessary in Roman society in the last decades of the Republic, when the economy and jurisprudence flourished. As a response to these growing demands, certain anonymous speakers created an elaborate system of prepositions in the common language. Through this way of giving information the already-existing tendency to pronounce the endings of cases in a lax manner became less troublesome. In relation to modern developments of common language Cicero and Caesar exercised restraint, following their ideal of elegantia. But they effected this restraint in different ways. This diversity in their use of prepositions - together with a variety of information about the history of Latin - gives us some hints in grasping the common language of that time. - Cicero's conservatism in the use of prepositions may be illustrated by his way of expressing the concept 'concerning', 'about'. Here he consciously avoided, like Terence, super, which was commonly used in this meaning, but he failed to notice that he used it three times in the Letters to Atticus. Besides de, he preferred the old prepositions in and ad. In this sense in also appears in a 'theme' outside the kernel of the sentence, e.g. o f f . 1,61. The origin of the meaning 'concerning' is to be found in local use (similar to other languages). De in the sense of 'concerning' is no exception; it is definitely not derived from the partitive use.

Personification is a process through which an inanimate entity appears as a fictitious character. This process might give rise to an allegory. It can be read in an either literal or figurative way. Personification indicates somebody's presence, whose only reality is its name, which is an abstract noun. Among those abstract nouns, only those indicating a favourably or unfavourably connotated notion can be personified. Through personification, the abstract noun loses its diathetic ambiguity and "referential dependance". It becomes a proper noun of a peculiar kind, for it is Justified. Personification does not transform abstract into concrete, but only what is abstract and inanimate into abstract and animated. It does not refer to any peculiar concrete experience, but only to a common cultural tradition.

Among the terms which mean “help” in Plautus’ corpus, copia occupies a characteristic place and can be understood either as a metonymy - when the servus callidus is called help - or as a metaphor - when the same slave is assimilated to the goddess Copia, who may have been worshipped in Rome -. These occurrences of Copia, which are always observed at very specific times in the plays, either at the beginning or at the end of a scene, are often emphasized by stylistic features such as sound effects, a picturesque style or puns on words. Moreover, such sentences, articulated on the word copia con- tribute to the “animation of inanimate things” which is so typical of Plautus’ art.

Though they have hardly been studied so far whether by ancient or modem grammarians, interjections, which make up a highly heterogeneous category, are items of discourse the use and stylistic value of which seems to depend on several factors. While the Ancients - among them Aulus Gellius - had already noticed that a particular interjection could only be used by men and some other by women, it has so far never been established that a writer or a literary genre could have an impact on the use of interjections. The present paper examines, on the one hand, the frequency with which interjections are used in a Latin corpus consisting of 1,600,000 words and. on the other, occurrences that specifically belong to prose, poetry or to this or that writer.

Cicero makes it plain in his remarks on translation that he holds lexical operations as an acceptable way of procedure for a competent translator (contrary to interpretes indiserti). This is supported by the methodological analysis of his translations. Comparing a prose translation and a poetical one, we can show the relative frequency of different lexical operations in their connection with the requirements of the genres, and with the predilections of the translator as well. We can find that these are used not only when the systemic differences between source and target language make them necessary, but they also become a means of aemulatio with the source text, which, by Cicero’s age, had become an essential component of artistic translation.

The study of tense variation in Latin texts can be renewed by using new devices, such as the lemmatized LASLA files and the Estela and Hyperbase programs (CNRS - UMR 6039 “Bases, Corpus et Langage”). Tense variation can therefore be related more precisely to various factors: writer, time period or literary genre. This will be exemplified by studying a corpus of classical historical texts, from Caesar to Suetonius. This paper will especially investigate the relations between tense variation and text sub-types, i.e. annalistic history, biography, military commentaries...

This paper deals with the syntactic and semantic properties of the so-called ‘hendiadyoin’ as well as with its value as a stylistic and rhetoric figure. Apart from its qualities as a specimen of pleonasm and abundancy, the hendiadyoin likewise serves as a tool to compensate the relative weakness of Latin to create nominal compounds, especially of determinative type. Casting a glance on the history of the Latin language and on the development of the individual style of some Latin authors, we can make remarkable observations. Whereas the figure is rather rare in early Latin, classical writers and poets give preference to it. Whereas, on the other hand, Cicero in his first texts prefers the combination of substantives with adjective attributes, he likes the hendiadyoin pattern very much in his mature speeches and dialogues.

This paper is dealing with particles such as at, autem, nain, sed. uero. operating in poetical context. These words have been previously described by Kroon 1995 in terms of ‘functional grammar’, so we use her work as a framework for our own insight in Catullus’poem, focusing on the rhythmical moves that particles contribute to provide. Often accounted as clitics, our particles may in fact be stressed by a discourse emphasis. Kroon’s description proves to be sound, even in poetical context, but, according to us, remains, somehow, too abstract, inasmuch as it does not explain the oral dimension of discourse organization.

This paper intends to show that, when a Latin word is provided with a great number of translations, distributed in several sections of an article in a bilingual dictionary, the real meanings of the word are far less numerous, and it often occurs that there is only one meaning. The Latin word receives various translations in modem languages because, from a language to another, the semantic features which compose the meaning of the word are differently distributed in the vocabulary. Our study is based upon the survey of iubeo, probo, arguo, uel, conscius, fides as examples. Some notions, like latent semantic features, or the organization of the numerous translations around one semantic feature as a common centre of the comprehension of the word (that centre being associated with various other semantic features in each particular translation), are used to provide difficult Latin words with a genuine unity of meaning.

This paper presents a method for measuring links between metre and syntax in the way three poets (Catullus, Horatius and Virgil) compose their hexameters. Based on the lemmatized files of LaSLa (Liēge) and on Tesniere’s dependency syntax, it aims to observe phenomena like (non-)coincidence of end of syntagm and end of verse line, syntactical consistency of the hexameter and so on. The test objects are: the place where sentences begin and end, syntactical connexions, syntagms exceeding the length of the line. It shows that poets like Catullus and Horatius don’t behave in the same way, and that the latter avoids less than the former this non-coincidence of meter and syntax: this is connected with a more “prosaic” nature of the writing of the Sermones.

The closing of an exchange requires cooperation from both speakers. When someone wants to end the dialog, whereas the other speaker doesn’t, conflict necessarily arises. The speaker reluctant to end the dialog is going to use every possible means to delay the closing, before the physical separation takes place. To prevent the closing, a speaker will usually use questions or orders, which compel the other to respond and prolong the exchange. He may also resort to summons (audin, sed quid ais) and addi- tions (thanks to the connective particles et, at and sed), or keep on speaking, without paying heed to the other speaker’s desire to leave. Whatever means the speaker may use to hinder the closing, the frequency of such passages in Plautus’ comedies is probably due to their comic effect.

This study stays at the crossroads of Rhetoric, Text Grammar and Textual Critic. It studies the division of chapters (made firstly by Gruter in 1607) of the Roman History (Historic Romwia: H.R.) of Velleius Paterculus: the old Rhetoric text types, the opening and closing marks, the characters and personages and the thematic sentences as cohesive devices. These thematic sentences are used as introductions of chapter, and they are followed by confirmatives with quippe and with demonstrative relatives. They constitute a mark similar to the propositions of the argumentative texts, with special syntactic characteristics in the text of Velleius. The studied corpus are the first 57 chapters of the second book of the Historia Romana.

On the subject of Plautus’s and Terence’s registers, dependent questions enable to draw a worthwhile comparison. For the most part, the subordinating terms as adverbs, pronouns, adjectives and particles are the very same in both authors. Most of the verbs governing the subordinate clauses are common between them. But the main point concerns the mood of the dependent verb : Plautus and Terence use equally indicative and subjunctive so that the difference between them is hardly noticeable.

Because of the interaction and oscillation between poetic imagery and subjective experience, Ovid’s exilic texts are a privileged locus for studying the relationship of rhetorical and poetic discourse. The use of repetition in its various features (at the level of the phoneme, of the word, of the meter, etc.) allows us to investigate the functioning of these two kinds of discourse in Tristia and Epistulae ex Ponto. In this paper we will examine the mise en scene of this rhetorical device in Tristia, considered as a structural part of a poetic design which includes it not as a mere formal technique but as a per- formative strategy. From this perspective, repetition assumes a functional role. For our analysis we will take some significative examples of a corpus which presents a semantic unity, the elegies of Naso’s journey to Tomi (7>. 1,2; 1,4; I, 10 and I, 11).