Ensaio teórico When the past becomes the present: the development of the old occitan passive voice* SLOBBE Bianca  revistadaabralin@gmail.com Universidade de Antuérpia 2 1 When the past becomes the present: the development of the old occitan passive voice 87 108 87-108 http://creativecommons.org/licenses/by/4.0/

This paper presents a generative syntactic analysis of the development of the passive present perfect in Old Occitan. In Classical Latin, the passive present perfect was expressed by a combination of a past perfect participle and a present tense form of BE. In Old Occitan, the same tense was expressed by a past perfect participle combined with a perfect tense form of BE. Giorgi & Pianesis' (1997) hypothesis, which accounts for this development by postulating the occurrence of a Perfect Shift in the development from Latin towards the Romance languages, is rejected in this paper. Instead, it is argued that the development described above was caused by a semantic change affecting the Latin verb form fuit in the Late Latin period, as a result of which verbal participle constructions were abandoned in favor of Small Clause structures.

 Resumo

Este trabalho é uma análise, do ponto de vista da sintaxe gerativa, do desenvolvimento do presente perfeito passivo em Occitano Antigo. No Latim Clássico, o presente perfeito passivo era expresso pela combinação de um particípio perfeito passivo com uma forma do verbo SER no presente. Em Occitano Antigo, o mesmo particípio perfeito passivo era combinado com uma forma do verbo SER num tempo perfeito. A hipótese de Giorgi & Pianesis (1997), que explica tal desenvolvimento postulando a ocorrência de uma Perfect Shift na transformação do Latim para as línguas românicas, é rejeitada. Em seu lugar, argumenta-se que as mudanças descritas acima foram causadas por uma mudança semântica que afetou a forma verbal latina fuit no Latim tardio, e que resultou no abandono das construções de particípio verbal em favor de estruturas de Small Clause.

gramática gerativa lingüística histórica sintaxe 1. Introduction

When we compare the Classical Latin passive present perfect (1) with its Old Occitan successor (2), a striking difference can be noted. In Classical Latin, a passive perfect participle (henceforward ‘p.p.p.’) was combined with a present tense form of BE. In Old Occitan on the other hand, the same p.p.p. was combined with a perfect tense form of BE. Judged by Classical Latin standards, this is one perfect tense feature too many. Still, (1) and (2) have exactly the same temporal interpretation.

In this paper I will argue that this development was caused by a semantic change affecting the verb form fuit, causing verbal participle structures to be abandoned in favour of Small Clause structures.

2. Theoretical Background 2.1 Semantic Background<bold id="bold-e84e5fc704a8b65337fe8b50ad64ef77"/>

I assume a revised Reichenbachian framework as proposed by Hornstein (1990)[1]. Reichenbach (1947)[2] postulated the existence of three temporal entities: Speech Time (S), i.e. the time at which the utterance is made, Event Time (E), i.e. the time at which the event or state denoted by the verb takes place and Reference Time (R), used to mediate in the relationship between S and E. Hornstein made several adjustments to this framework, one of which concerns the relationship between S, E and R. For Reichenbach, this was a ternary relationship. Hornstein claimed that this relationship should be split into two binary relationships: one between S and R and one between R and E. Thus, the relationship between S and R determines whether an event or state is located in the past, present or future: in the present tense, Speech Time and Reference Time coincide (S,R); in the past tense, Reference Time precedes Speech Time (R_S) and in the future tense, Speech Time precedes Reference Time (S_R). The relationship between E and R determines whether an event or state is interpreted as perfective, neutral or prospective. If E precedes R (E_R), the verb has a perfective reading; if E and R coincide (R,E), the event or state is interpreted as neutral and if R precedes E (R_E), the reading is prospective. Thus, for example, the simple present is represented by (S,R,E), the simple past by (E,R_S), the present perfect by (E_S,R) and the past perfect by (E_R_S).

 2.2 Syntactic Background

I adopt the Minimalist framework as proposed in Chomsky (1995)[3]. I follow Giorgi & Pianesi (1997)[4] who postulate the existence of two tense projections, T1P and T2P. According to them, T1 realises the relationship between S and R, thereby determining whether the event or state is located in the past, present or future, while T2 realises the relationship between R and E and thus determines perfectivity. Note that these tense projections are only realised in the structure when the temporal entities do not coincide. This means that the past perfect, for instance, has both a T1 and a T2 projection whereas the present perfect only has a T2 projection and the simple present has neither. In this paper, I would like to focus attention on the behaviour of T2. I will therefore concentrate on examples in which R and S coincide and, consequently, T1 is absent.

 3. Terminology and Definitions

At this point some clarification concerning the terminology used in this paper seems in order. Nedjalkov (1988) distinguishes three basic types of predicates: actions, states and qualities. I will leave qualities aside and focus on actions and states. According to Nedjalkov, states may be subcategorised into natural or primary states and secondary states. Natural states may come into being by themselves, while secondary states are always the result “of somebody’s conscious action or activity”. In this paper, I will use the term ‘state’ to refer to Nedjalkov’s secondary states, with one crucial modification: although they are the result of a previous action or event, this does not necessarily have to be the conscious action of a rational agent. (A door, for example, may be opened by a gust of wind.) With the same reasoning I have replaced Nedjalkov’s ‘action’ with ‘event’, because ‘action’ seems to imply the presence of a conscious or at least animate agent, which is not always the case.

 4. Data and phrase structures 4.1. Overview of the data

The diagram below gives a schematic overview of the data to be discussed in this paper.

4.2. Data: Classical Latin

In Classical Latin, the passive simple present was represented by a synthetic form, consisting of the present stem and a passive ending (3). This verb form denoted an uncompleted event, concurrent with the Time of Speech, as exemplified in (4).

As mentioned earlier, the passive present perfect was formed analytically, combining a p.p.p. with a present tense form of BE, esse (5). This verb form had two possible interpretations: it could denote (a) a state concurrent with Speech Time as the result of the completion of the event in the past (exemplified in (6a)) or (b) a completed event, located in the past with respect to Speech Time (exemplified in (6b)).

For resultative verbs, the p.p.p. could also be combined with a perfect tense form of esse (7), denoting a state located in the past with respect to Speech Time as the result of the completion of the event in a further past. Crucially, the state has ceased to exist at Speech Time. This is shown in example (8). This construction could not have an eventive reading.

4.3. Phrase structures: Classical Latin

Before moving on to later data, let us consider the phrase struc-ture of participle constructions in Classical Latin.

Giorgi & Pianesi (1997)[4] assume a verbal structure as in (9), which I have implemented in (10):

According to this analysis, the participle starts out as the head of a VP and then moves on to T2. T1 is realised by an auxiliary. If T1 is neutral, the auxiliary has present tense by default. This has an important consequence. As T2 is occupied by the participle, the auxiliary may only bear T1 features. In other words: according to this derivation, participle constructions can only be formed with non-perfect tense auxiliaries. It is therefore unable to account for the construction fuit apertum. Furthermore, it cannot explain why fuit apertum should only yield a state reading, while est apertum can yield both a state and an eventive reading. This structure seems, therefore, too restrictive.

Kayne (1993)[5] proposes that the past participle of Italian unaccusatives should be regarded as an adjective, selected by the copula BE. This assumption leads us to assume a Small Clause structure as in (11), implemented in (12). In this structure, the participle is located in a Small Clause, selected by a copula. This copula may project both a T1 and a T2 projection.

Unlike the verbal structure, this structure accommodates forms like fuit apertum as well as est apertum. However, just like the verbal structure, it cannot account for the difference in semantic distribution between fuit apertum and est apertum. I propose that the verbal structure as given in (9) and the Small Clause structure as given in (11) are complementary and are both needed to account for the Latin data. Let us assume that the semantics are connected to the syntax. I propose that Small Clause structures will yield adjective-like, i.e. state readings whereas verbal structures will yield verbal, i.e. eventive readings. This assumption leads to two predictions: (I) Est apertum should be ambiguous. As it fits structure (9) as well as (11a), it should be able to denote a completed event anterior to Speech Time or a state concurrent with Speech Time. (II) Fuit apertum should only be able to refer to a state anterior to Speech Time, as it is only accommodated by structure (11b). This is precisely the situation in Classical Latin.

 4.4. Data: Late Latin

A source that provides interesting data concerning the passive paradigm in Late Latin is the so-called Peregrinatio Aetheriae. This text, which is traditionally dated around 384 AD and ascribed to a nun from the Occitan area, can be divided into two separate parts that differ substantially when it comes to syntax and vocabulary. I believe that the two parts were written by different authors and represent different stages in the development of the Occitan passive voice.

 <italic id="italic-60a3b22db7aeafdd75f9db145008664c">4.4.1. Peregrinatio Aetheriae 1</italic>

The first part of the Peregrinatio (chapters 1-23) displays almost the same use of tenses as Classical Latin did. In (13) an uncompleted event concurrent with Speech Time is represented by a synthetic present passive (cf. apertur in (3)).

Just like in Classical Latin, the passive present perfect, consisting of a p.p.p. and a present tense form of esse (cf. est apertum in (5)), can denote either a state concurrent with Speech Time or a completed event located in the past with respect to the Time of Speech. This is shown in examples (14a) and (14b) respectively.

However, the interpretation of the construction fuit apertum, combining a p.p.p. with a perfect tense form of esse (cf. (7)), has changed. As example (15) shows, this construction now accommodates an eventive reading, contrary to the situation in Classical Latin. In this reading, it denotes a completed event, located in the past with respect to Speech Time. It is thus synonymous with the eventive interpretation of est apertum as given in (6b) and (14b).

At this point, an additional note concerning the construction fuit apertum seems necessary. In the Peregrinatio there are no attestations of this construction denoting a state anterior to Speech Time as it did in Classical Latin and as it should do according to my analysis. However, this interpretation is not realised by any other construction either; it is simply absent in this text.

This text can therefore provide no evidence as to whether the construction fuit apertum could still denote an endstate anterior to speech time. I have found occurrences of this construction with the state reading with other authors, whose texts display the same paradigm as the second part of the Peregrinatio. These authors will be discussed in section 3.3.2.

As of yet, I have not found other texts with the same passive paradigm as the first part of the Peregrinatio. Therefore, it is impossible to determine whether the construction fuit apertum could have a state reading at this stage. However, because it did have that interpretation in Classical Latin, Late Latin 2 and Old Occitan, I assume this is a continuous situation and the absence in the Peregrinatio is merely an accidental gap. I therefore inserted this form in brackets under Late Latin 1 in the diagram.

 4.4.2. <italic id="italic-fce93eb45f144023b2b1786c4f256bb7">Peregrinatio Aetheriae 2</italic> and other authors

The second part of the Peregrinatio (chapters 24-49) shows the same development and ads another: the analytic construction est apertum may now be interpreted in the same way as the synthetic construction apertur, denoting an uncompleted event concurrent with the Time of Speech. This is clearly shown by example (16), where ornatus est is modified by the adverb cotidie and coordinated with the synthetic present passive celebratur.

Three other authors from this area, namely Phoebadius Aginnensis (ante 392 AD), Prosper Aquitanus (floruit 420 – 425 AD) and Salvianus Massiliensis (floruit 435 – 440 AD), display the same passive paradigm in their texts. This suggests that the second part of the Peregrinatio too should be dated around the beginning of the 5th century.

Examples (17) to (22) illustrate the passive paradigm of Prosper Aquitanus: the synthetic passive simple present is still in use (17); est apertum may refer to an uncompleted event concurrent with Speech Time (18), a state concurrent with Speech Time (19) or a completed event anterior to Speech Time (20); fuit apertum may denote either a completed event anterior to Speech Time (21) or a state anterior to Speech Time (22).

4.5 Data: Old Occitan

Old Occitan takes the development that has been started by the Late Latin authors even further. The synthetic passive present (apertur in Latin) is abandoned in favour of the analytic form es ubertz (23), combining a p.p.p. with a present tense form of BE, similar to the Latin est apertum (5). This form now denotes an uncompleted event, concurrent with the Time of Speech or a state concurrent with Speech Time as the result of the completion of the event in the past.

The ‘completed event in the past’-reading of est apertum, which was available in Classical Latin (cf. (6b)) and Late Latin (cf. (14b) and (20)), is lost in Old Occitan es ubertz. This interpretation is picked up by fo ubertz (24), the morphological equivalent of the Latin fuit apertum (7), combining a p.p.p. with a perfect tense form of BE. Fo ubertz also continues to denote a state that existed in the past as the result of the completion of an event in a further past, but no longer exists at Speech Time.

4.6. Summary

Summarising these data, we can say that we are dealing with a development that takes place in three subsequent phases. In the first stage (Late Latin 1), the construction fuit apertum, originally only used to denote a state anterior to Speech Time, starts to allow an eventive reading as well, thereby competing with the construction est apertum. In the second stage (Late Latin 2), est apertum starts to compete with apertur, both denoting - among other things, in the case of est apertum - an uncompleted event concurrent with the Time of Speech. In Old Occitan, the battle appears to have been won by the “invading” constructions: es ubertz has lost its anterior-eventive reading and apertur has been abandoned altogether.

 5. Hypothesis: expansion of Small Clause structures 5.1. Problems posed by Late Latin and Old Occitan

The Late Latin and Old Occitan data discussed above pose several problems to my analysis of participle constructions as given in section 3.2. According to this analysis, fuit apertum should not be able to yield eventive readings, as it does in Late Latin and Old Occitan. Second, in the eventive reading, est apertum is expected to denote a completed event anterior to the Time of Speech, not an uncompleted event concurrent with Speech Time as it does in Late Latin and Old Occitan. Furthermore, the disappearance of the synthetic present passive needs to be accounted for.

 5.2. The Perfect Shift-hypothesis rejected

Giorgi & Pianesi (1997)[4] postulate the occurrence of a so-called ‘Perfect Shift’ in the development towards the Romance languages, causing the perfectivity features to shift from T2 to T1. At first sight, this seems to account for the change in use of tense on the auxiliary: if T2 no longer checks perfectivity features on the participle, these features need to be realised by the auxiliary. This implies, however, that the auxiliary may at most have present perfect tense and not, for example, past perfect tense, as T1 can only check one set of features at a time. That is, unless we allow a functional projection to check more than one set of features, which seems undesirable. Example (25) shows that a participle could be combined with an auxiliary with past perfect tense in Late Latin, proving that the Perfect Shift-hypothesis cannot be upheld in this case.

5.3. Expansion of Small Clause structures due to a semantic change

I believe the development discussed in this paper was caused by a semantic change that affected the verb form fuit in itself. In Late Latin 1, it can be used as the perfect tense of a verb of motion, as is shown in (26).

If a motion verb enters a Small Clause structure, it can easily yield an eventive reading. This often happens in Dutch, as illustrated by (27).

I propose this is what caused the development in the first part of the Late Latin 1: when the existential variant of fuit entered a Small Clause structure, it yielded a state reading; when the motion verb fuit entered a Small Clause structure, it yielded an eventive reading.

In Late Latin 2, fuit can no longer denote movement when it is used independently. A sentence like example (26) does not occur in this part of the text. But fuit maintained its eventive reading in participle constructions. I assume it was re-analysed as a perfect tense form of esse and the eventive interpretation was transferred to the existential fuit. As a result, its eventive reading could be extended to Small Clause structures with a present tense form of esse. This is what caused the change in interpretation of est apertum.

Once the new interpretations of the Small Clause structures had been accepted, the verbal structures could be abandoned in order to create a coherent, non-redundant paradigm. This is what happened in Old Occitan.

 6. Conclusion

In this paper I have discussed the passive voice of Classical Latin, Late Latin and Old Occitan. I have argued that Classical Latin had a strict passive paradigm, using both verbal participle structures and Small Clause structures with a clearly distinct domain. I have furthermore argued that the change in the use of tense on the auxiliary in the development towards the Old Occitan paradigm was started by a semantic change affecting the verb form fuit. Interpreted as a verb of motion, it accommodated eventive readings of Small Clause structures with a perfect tense copula (Late Latin 1). After having been reanalysed as a verb of existence, it caused the eventive reading to expand to Small Clause structures with a non-perfect tense copula (Late Latin 2). Eventually, Principles of Economy caused the verbal constructions to disappear (Old Occitan).

 7. Abbreviations

The following abbreviations have been used in this paper:

* A slightly different version of this paper, entitled 'Doubling the Tenses: the Development of the Old Occitan Passive Voice' has been accepted for publication in Rivista di Grammatica Generativa, Proceedings of the xxix Incontro di Grammatica Generativa part II. The present version has benefited from the comments provided by the reviewing committee of ABRALIN.

A slightly different version of this paper, entitled ‘Doubling the Tenses: the Development of the Old Occitan Passive Voice’ has been accepted for publication in Rivista di Grammatica Generativa, Proceedings of the xxix Incontro di Grammatica Generativa part II. The present version has benefited from the comments provided by the reviewing committee of ABRALIN.

Cf. Pinkster (1987). Strictly speaking, the correct Classical Latin form would be apertum est, with the auxiliary following the participle. In the present paper, I abstract away from word order differences between Classical Latin and its successors. I take these differences to be the result of the general shift from SOV- to SVO-surface order, and therefore unrelated to the issue under discussion here.

Cf. Grandgent (1905)[6]

With ‘Classical Latin’, I refer to Latin as it was written in the first century B.C.

For Italian, this derivation is rejected by Cocchi (1994)[7]. I am convinced however, that we need this construction to account for the Latin data.

Note that the English translations might be misleading: in (6b) a present perfect is used, while in (14b) a simple past is used. This difference is caused by the fact that in (6b), the matrix verb has present tense while in (14b) the matrix verb has simple past tense. In that case, English requires a simple past tense in the subordinate clause as well. Latin doesn’t. In latin, the morphological tense and temporal interpretation of facta sunt (6b) and incensus est (14b) are exactly the same. This problem concerning the clarity of the translations will continue to occur throughout this paper. I will not mention it any further.

If laetatus sum were to have an eventive reading, the agent of the event would have to be his quae dicta sunt mihi. The preposition in clearly shows, however, that his quae dicta sunt are not the agent; otherwise the preposition a / ab (‘by’) would have been used.

Anglade (1977)[8], Grandgent (1905)[6], Mann (1886)[9], Paden (1998)[10].

Anglade (1977)[8], Grandgent (1905)[6], Mann (1886)[9], Paden (1998)[10].

This change affected all forms of the present perfect and pluperfect of this verb, both in the indicative and the subjunctive (perf. fui, fuisti, fuit, fuimus, fuistis, fuerunt, fuerim, fueris, fuerit, fuerimus, fueritis, fuerint, pluperf. fueram, fueras, fuerat, fueramus, fueratis, fuerant, fuissem, fuisses, fuisset, fuissemus, fuissetis, fuissent, fuisse). For practical reasons, I use the 3rd person singular of the present perfect indicative to refer to all these forms.

An anonymous reviewer has pointed out to me that the verb form fuit is derived from the Indo-European aorist, which had an eventive reading. In that case, the possibility to use fuit as a verb of motion is a result of its eventive reading, not the other way around as is suggested in the text. In that case, the question rises why fuit did not have an eventive reading in Classical Latin. This issue is left for further research. For the moment, it suffices to conclude that the ability of fuit to yield an eventive reading when combined with a past participle in Late Latin 1 is linked to its ability to be used as a verb of motion.

 Montpellier CEO 1976 ALIBERT Loïs Grammatica occitana segon los parlars lengadocians Paris Éditions Klincksieck 1977 ANGLADE Joseph Grammaire de l’Ancien Provençal ou Ancienne Langue d’Oc. Phonétique et Morphologie. 1996 CETEDOC Brepols Pub Cetedoc Library of Christian Latin Texts volume 3 (cd-rom) MIT Press 1995 CHOMSKY Noam The Minimalist Program 2 3 6 1994 COCCHI Gloria Probus ‘An explanation in the split in the choice of perfect auxiliaries’ KUL Leuven 1991-1992 D’HULST Yves La sintassi del passivo italiano nel quadro della grammatica universale Oxford University Press 1997 GIORGI Alessandra PIANESI Fabio Tense and Aspect. From Semantics to Morphosyntax Boston 1905 GRANDGENT C H An Outline of the Phonology and Morphol-ogy of Old Provençal Heidelberg 1929 HERAEUS W Silviae vel potius Aetheriae peregrinatio ad loca sancta (Itinerarium Egeriae). MIT Press 1990 HORNSTEIN Norbert As Time goes by. Tense and Universal Grammar 47 1993 KAYNE Richard S Studia Linguistica "Toward a Modular Theory of Auxiliary Selection" Marburg 1886 MANN Paul Das participium praeteriti im Altprovenzalischen Muiderberg 1977 MOK Q I M Manuel pratique de morphologie d’Ancien Occitan 1987-1991 PACKARD HUMANITIES INSTITUTE Los Altos (California) PHI Latin Texts cd-rom volume 5.3 New York MLAA 1998 PADEN William D An Introduction to Old Occitan New York Collier-Macmillan 1947 REICHENBACH Hans Elements of Symbolic Logic Aix-en-Provence Edisud 1995 SAUZET Patrick UBAUD Josiane Le Verbe occitan. Lo Vèrb occitan Proceedings of the xxix Incon-tro di Grammatica Generativa, Urbino Padova February 2003 SLOBBE Bianca Rivista di Grammatica Generativa, Unipress 'Doubling the Tenses: the development of the Old Occitan passive voice'