CHANGE. FROM LATIN MIDDLE TO ROMANCE REFLEXIVE
MARIA M. MANOLIU
Abstract. Whether they were traces of an ancient merger or just an economical tool for expressing the lack of interest in speaking about the Agent, the non-active Latin forms – otherwise called “medio-passive” – constitute another example of how interpretations governed by the grammar of contemporary languages can lead to an inappropriate description of a dead language. If there has been no much discussion about the interpretation of the active voice, the forms in -R have been subjected to a variety of interpretations mostly guided by translations in a specific contemporary language. One might think that looking for an invariant semantic feature capable of explaining all these derived values, contextually determined by the verbal stem, is an exercise in futility.
However, in our opinion, it is at least worth trying to find a link between all these multiple functions encoded in a single morphemic paradigm. But before presenting our arguments in favor of a unifying discourse model, we shall examine the positive trends as well as the weaknesses of a few models of the category of voice in general, and of Romance voice in particular.1
1. A FEW BASIC CONCEPTS
1.1. Cognitive Grammars
According to Langacker's cognitive grammar, the passive imposes a choice of the trajector that reverses the relation between the figure and the ground of the active construction. (1982, 1987: 351–352). The figure is “the structure perceived as
‘standing out’ from the remainder, which is represented by the ground” (Langacker 1987: 120).
The function of the past participle in English is two-fold:
“(i) it suspends the sequential scanning of the verb stem, converting it into a complex atemporal relation (hence the temporalizing be is required in a finite clause);
1 If there seems to be a consensus about the fact that the term diathesis corresponds to the semantic area expressed by voices (a universal feature of language), the definition of voice as a grammatical category (language specific) is not without its controversial aspects.
RRL, LI, 2, p. 245–266, Bucureşti, 2006
and
(ii) it imposes a figure/ground alignment distinct from that of the stem, specifically with respect to the choice of trajector.” (see Langacker 1987: 351).
1.2. Discourse factors
Touratier (1984: 90) stresses that voice is the effect of a structural reorganization of the sentence which corresponds to a reorientation at the level of meaning:
Mais du fait de la suppression du premier argument et de la promotion structurale du second argument à la fonction de sujet, il y a forcément une réorganisation de cette signification. Celle-ci est orientée différemment: son contenu, inchangé, est rapporté au second argument et non plus au premier argument, qui ayant été supprimé ne peut plus être pris en compte.
In his OT model, Sells (2001) introduces topicality beside the role of participants, agent and patient, in order to calculate the possibilities that voices can encode in any language. In fact, Sells (2001: 360) speaks of a scalar topicality:
Scalar Topicality is determined by the discourse measures of Referential Distance and Topic Persistence. Referential Distance is the measure how far back in discourse the previous mention of a referent is, from a given point;
the lower the measure of Referential Distance, the more topical entity is.
Topic Persistence measures how many times in succeeding discourse a referent is mentioned, from a given point; hence the higher the measure, the more frequent and topical the entity in question is.
The possible combinations of these parameters may be summarized as follows (Sell 2001: 363):
Voice Type equivalent Expression Linking LRS
Active a1-p1 aS1-pO1 AP
Inverse–GR a1-p1 aO1-pS1
Antipassive a1-p0 aS1-pOBL0 Ap
Passive a0-p1 aOBL0-pS1 aP
Super Active a2-p1 aS2-pO1
Super Inverse-GR a1-p2 aO1-pS2 Super Antipassive a2-p0 aS2-pOBL0
Super Passive a0-p2 aOBL0-pS2
Lacking any corresponding examples, this calculus fails to carry conviction, especially since, as will be shown below, this model, which takes into account only two roles, the Agent and the Patient, cannot account for languages which encode grammatically the difference between constructions with a subject referring to an
Agent and those with a subject referring to an Experiencer. This is the case with Latin forms in –r and the Romance reflexive which can have an Experiencer as their subject as distinct from active constructions (see 1).2
(1) laetor ‘I am happy’, ‘I rejoice’; doleor ‘I suffer’
Fr je me réjouis, Sp. me alegro, It. si rallegra, Rom. mă bucur
‘I rejoice/ I am glad’.
2. SYNCHRONIC VIEW OF LATIN VOICES
Everybody agrees that, in Latin, the category of Voice had only two members:
(i) Active and (ii) the so-called Medio-passive.
2.1. The values of the active voice
The active voice was agent-centered. It represented the neutral term of the category, since it occured also with non-agentive verbs, such as esse ‘to be’, habeo ‘to have’, etc.
In archaic Latin the active voice could be used also with verbs expressing
‘personal care’:
(2) lavo ‘I wash myself’
When co-occurrent with the inchoative suffix -sc- it could be the expression of a change of state:
(3) dulcescat ‘[it] turns/ becomes sweet’
2.2. The functions of the forms in –R
According to classic descriptions, often conditioned by translations in a variety of contemporary languages – either Romance or Germanic – the values of the forms in –R are characterized by a mosaic-like variety:
(a) impersonal (with either transitive or intransitive verbs):
(4) dicitur ‘they say, it is said’; itur ‘they go’ (Ernout-Thomas, 1953: 204-205);
(b) middle-reflexive (when the subject shares the roles of both an Agent and a Patient):
(5) lavari, tergeri (Poen. 219) ‘to wash, to clean [oneself]’ (intrinsic passive);
(c) middle/ active (especially verbs of feeling) (the subject refers to an Experiencer):
2 In fact, Romance syntax is still sensitive to the difference between Agent and Experiencer:
see, for example, constructions in which the topical personal Experiencer is either in the dative or in the accusative: Rom. mi-e frig lit ‘to me is cold’ i.e. ‘I am cold’, mi-e foame ‘to me is hunger’ i.e. ‘I am hunger’, lit. me:ACC hurts head’ i.e. ‘my head hurts’, Sp. me duele la cabeza lit. ‘to me hurts the head’, i.e. ‘my head hurts’, etc. Cf. also Lat mihi placet – Sp. me gusta, It. mi piace, Rom. îmi place lit. ‘to me pleases’, i.e. ‘I like...’
(6) laetor ‘I rejoice’, miror ‘I am amazed [at], I am surprised, I admire’, queri ‘to complain, to lament’, cf. also the semi-deponent gavisus sum ‘I rejoiced’ and late Lat. doleor ‘I suffer’ (CIL, 6. 23176; CIL 02, 03249 = CILA-03-1, 045.
CIL 02, 00041 (AE 1982, 0489).
(d) inchoative (the subject is the Locative/ the location of the change and of the Object that changes):
(7) ignis exstinxitur ‘the fire extinguished [by itself]’;
(e) verbs of change in position:
(8) ferri ‘to rush, to flee’, moueri ‘to move [oneself]’, uehi ‘to transport [oneself]’, uerti ‘to turn around’, etc.
(f) passive:
(9) amor ‘I am loved’
(g) factitive (the subject refers to the Patient acted upon):
(10) aduruntur (Tusc.5, 27, 77) ‘they let themselves be burnt’ (cf. Fr. ils se laissent brûler – Touratier, 1984:81)
(h) deponent:
(11) imitor ‘I imitate’; sequor ‘I follow’.
(i) reciprocal:
(12) copulari dexteras ‘to shake [the] right [hands]’.
For more than twenty years – due to the preponderance of syntactic models in linguistics – descriptions of the Latin forms in -r have been concerned only with the syntactic characteristics of the forms under discussion. Consequently, the medio- passive forms have been categorized as means of intransitivizing the verb, in other terms, of reducing the number of verb-valences. If it is true that the impersonal and the passive constructions can be used for demoting the Agent from the core structure of the sentence (either by not mentioning it at all or by presenting it as a focalized prepositional complement), the syntactic hypothesis that the medio-passive forms were a means of intransitivizing the verb (see Touratier 1984, or Everaert 1986, for the generative treatment of Romance reflexive constructions; Italian; Romanian) cannot account for the behavior of all forms in -tur. Many Latin verbs, including the deponents, were transitive. For example, dexteras ‘[right] hands’ is the object of copulari in (12) and ferrum ‘the sword’ is the direct object of cingor in (13): 3
3 See also other transitive medio-passive verbs: conspicor ‘to spot, to see [something], to catch sight of’; aemulor ‘to emulate [somebody], to rival’; aggredior ‘to approach, to address, to attack’;
sequor ‘suivre [sombeody]’, etc. A verb such as suggredior changes its meaning according to its transitive or intransitive use, namely: when transitive, suggredior means ‘to board a ship, to tackle a question’ – cf. fr. ‘aborder [qc]’; when intransitive, it means ‘to get close to something’ – cf. fr.
‘s’approcher [de qc]’.
(13) ferrum cingor sword:ACC fasten:I
‘I fasten my sword’
As I have already emphasized, it is rather hard to believe that a single morphological form could express so many and even contradictory values. I am rather inclined to assume that it is possible to find an invariant that can account for all these uses. I shall therefore argue in favor of the hypothesis that Latin voice encoded the relation between Agency and discourse salience. It has to be noted at this point that Agent refers here to a doer (in Meillet’s words, une ‘force agissante’), not necessarily a person or a living being.4
In accordance with its transparent label, the active voice was typically agent- centered, in the sense that its subject, typically referring to an active force, was the speaker’s focus of attention, whereas the medio-passive form in -r signaled the fact that the doer was not the most salient discourse entity; in brief, the medio-passive was mainly event-centered.
2.2.1. Arguments against a passive interpretation. Even according to classical grammarians, the impersonal interpretation (where the agent is not explicitly present) was more important than the passive value. For example, according to Meillet &
Vendryes (1960: 324):
le sens impersonnel marquant simplement que l'action est en voie d'accomplissement ou accomplie (suivant qu'il s'agit de l'infectum ou du perfectum) domine la valeur du passif latin
According to Ernout and Thomas (19531 - 1993: 206), the distinction between active constructions and the personal passive is secondary, since the passive does not necessarily imply that the subject is the undergoer:
Seuls les verbes transitifs ont en général un passif personnel. Mais l’équivalence établie habituellement entre la tournure active (me diligit pater) et la tournure passive (a pater diligor) est imparfaite et secondaire. Le passif personnel, en effet, n’implique pas nécessairement que le sujet subit l’action. Souvent il garde la valeur de l’impersonnel, désignant un état ou une action indépendamment de tout sujet determiné, ce qui en fait aussi une expression de l’indéfini: Pl., Cas.121: dabitur tibi amphora ‘on te donnera une amphore’ (la question de savoir qui la donnera n’intéresse pas); Pl. Mi. 674: quod sumitur ‘ce qu’on
4 See Meillet, 1921.1: 199-229 and 1937.2: 24 – 28. In Foley and Van Valin (1984: 290-300) the doer is the argument of a predicate that corresponds to the participant who performs, effects, instigates or controls the situation denoted by the predicate. The opposite term is the undergoer, which corresponds to the participant who does not perform, initiate or control any situation but rather is affected by it in some way. For details on the inherent seme [Incapable of being an active participant], see Manoliu 2006.
dépense’, en face de v. 673: sim quid sumas (2e pers. Indéf.). Cette valeur peut êrre sensible même à une 1re personne: Tér., Ad. 911: iam lepidus uocor ‘voici qu’on m’appelle charmant’.
Tous les verbes transitifs, du reste, n’ont pas de passif. Alors, c’est un autre verbe qui en tient lieu: fio, par ex. pour facio, sauf faciendus et factus; disco ‘je suis instruit’ pour doceo ‘j’instruis’ (doceor est postclassique); intereo ‘je peris’
pour interficio ‘je tue’; [...]. Des périphrases servaient de passif au déponent:
usui esse ‘être utile, utilisé’, en face de uti ‘utiliser’; admirationem habere (mouere) ‘être admiré’, en face de admirari ‘admirer’.
A sentence such as (14):
(14) cum a Cotta resisteretur (Caesar, BG, 5.30,1), as by Cotta resist: medio-passive.past.3rd means something like:
(14a) ‘as there was resisting by Cotta’
(in Meillet's and Vendryes's words, ‘puisqu'il y avait de la resistance de la part de Cotta’), rather than:
(14b) ‘as it was resisted by Cotta’.
The translation of (15) proposed by Touratier (1984: 89) where the by + NP construction is translated by a French subject does not correspond to the discourse organization of the Latin construction (see 15a):
(15) ab hostibus constanter pugnabatur (Caesar, BG, 3.25,1) (15a) ‘Les ennemis se battaient avec ténacité’.
i.e. ‘The enemies fought with tenacity’
In our opinion, a translation in Meillet’s or Guillaume’s terms would be more appropriate (see 15b):
(15b) Quant aux ennemis, il y en avait un combat constant [de leur part]
‘As for the enemies, there was a permanent fight on their part’
According to classical grammars, in Latin there was no real agentive complement; the forms in -r could be determined by a dative of the author or by a cause complement (see 2.2.2 below). But, if the forms in -r cannot be considered as expressions of a true passive, what functions did they fulfill?
2.2.2. Arguments for a middle interpretation. In our opinion, the forms in -r seem more closely related to the I.E. middle voice than to the passive. Our hypothesis rests upon the following interpretations of the functions and values of the Latin forms in -r. According to Ernout, Thomas (19531 - 1993: 2007-8), the characteristic function of the Latin passive construction was to stress (to set in a prominent position) the
verbal notion rather than the agent (in their own words: de mettre en relief la notion verbale plutôt que l’agent):
L’ablatif précédé de a ou ab est encore rare en v. latin. Il ne devient plus fréquent qu’à la fin de la période républicaine: le sens de la préposition est alors affaibli, et l’indication de l’origine ou de notions voisines tend à s’effacer.
Cependant même ainsi ce tour n’est pas toujours l’équivalent exact de l’actif avec sujet au nominatif. On y retrouve souvent la fonction propre au passif de mettre en relief la notion verbale plutôt que l’agent [our emphasis]: Tér., Ph.
854: sine controuersia ab dis solus diligere ‘tu es sans conteste le seul chéri des dieux’; César., B.G. 2, 26, 3: cursus incitato in summo colle ab hostibus conspiciebantur ‘ayant pris le pas de course, ils étaient aperçus de l’ennemi au sommet de la colline’; dans le premier passage c’est l’idée d’une affection unique, dans le second, celle d’une apparition soudaine qui sont ainsi soulignées.
Guillaume (1971, 2: 189).defines the “impersonal meaning ” of the middle in the following way:
Dans les langues anciennes le moyen, plus ou moins infléchi en direction du passif, fait souvent l'objet d'une évasion dans l'impersonnel, selon laquelle l'incidence active maintenue dans le verbe ne trouve pas de support personnel, ce qui conduit à faire, dans le cadre de la voix moyenne, le verbe sujet, en quelque sorte, de lui même. C'est ainsi que pugnatur signifiera ‘il est combattu, il y a combat’
For Flobert (1975: 523), les déponents marquent que le sujet est profondément impliqué dans le procès, affecté par lui; l'actif est plus neutre, plus banal et le sujet y est moins engagé. Quant au passif, l'agent manque ou ne figure qu'à titre subsidiaire, [...] les morphèmes passifs transforment complètement la représentation du procès.
According to Kemmer (1993, Chapter 4), the middle is characterized by a low participant distinguishability (in her words, by a low degree of elaboration of events, subsuming low participant and sub-event distinguishability).
It is also interesting to recall here the relation between the forms in -r and the so-called internal diathesis, defined as expressing the fact that the subject is affected by the activity. As Joffre (1995) puts it, the basic meaning of the forms in -r is internal diathesis (la diathèse interne): the two concepts (nominal and verbal) are superposed (les deux concepts (nominal et verbal) se superposent, ils sont impliqués l'un dans l'autre (p. 236). See also:
les deux notions sont étroitement associées, leur adéquation réciproque est signalée (p. 195).
In the case of impersonal constructions:
le morphème -tur modifie le contenu lexical du verbe en lui ajoutant un sème qui souligne l'existence du procès (p. 193).
When analyzing the deponents or the passive, she stresses that on note toujours une implication du sujet dans le procès – p. 235. In our opinion, in spite of the fuzzy terms (concepts that are superposed, implication of the subject in the process, etc.), Joffre’s observations show that there is a close link between the subject and the predicate, which in fact fits in with our idea that both the subject and the predicate have the same degree of discourse prominence.
All these remarks point to the hypothesis that, in medio-passive constructions, the linguistic entity referring to the action (read the event) and not the subject had the highest discourse salience. These interpretations of the Latin medio-passive support our hypothesis, that, as we have already emphasized, Latin diathesis (the semantic area expressed by morphologically or syntactically marked voices) was organized around the agency, interpreted as the capacity of influencing human life (which is not the same as the feature animate referring to the property of living). The active voice was typically centered on the agent, the one that had the power of affecting human life in positive or negative ways. On the contrary, the form in -r was the expression of a linguistic reconstruction centered on the event as a whole. The difference between active and medio-passive may be roughly described as follows: active constructions encoded events when the speaker talked about the fact that an active entity acted upon somebody/something whereas medio-passive constructions encoded events when the speaker wanted to talk about the fact that something happened to somebody/ something.
3. THE MIDDLE VOICE AND LATIN REFLEXIVE PRONOUNS In our opinion, in classical Latin, when the reflexive pronoun co-occurs with the active form, it overtly expresses a reduplication of the subject, whereas the forms in – R can carry the conventional implicature that the subject refers to two Roles and that the Agent does things deliberately. 5 According to Flobert (1975: 387):
Il s'opère alors un véritable dédoublement du sujet tout à la fois agissant et agi;
le procès reflète une volonté délibérée de soi sur soi: le reflexif est donc dualiste.
(16) se occidere
‘to kill onself’
5 For the concepts of conversational and conventional implicature see Levinson (1983, 2000) and, more recently, the survey of various positions concerning Grice’s ‘Principle of Cooperation’ in Jaszczolt (2002: 207-223). There are a kind of non-truth conditional inferences that are not derived from superordinate pragmatic principles like the maxims, but are simply attached by convention to particular lexical items or expressions. Properties: they are detachable – they depend on the particular lexical items used (in the present case: the reflexive clitics); they are not cancelable, because they do not rely on defeasible assumptions about the nature of the context; they will not be calculated using pragmatic principles and contextually dependent knowledge, but rather given by convention. The items carrying conventional implicatures do not seem to have radically different interpretations in different contexts.
(17) mulier quae se suamque aetatem spernit, Plaut. Mo 250
‘woman which rejects/hinders herself and her own age’
(18) ubi se adiuvat ibi me adiuvat, Per. 304
‘where he helps himself there he helps me’
With verbs of movement (such as se mouere), the reflexive pronoun
souligne que l’initiative du mouvement appartient au sujet, tandis que le passif intrinsèque moueri ‘se mouvoir’ marque seulement l'implication du sujet dans un mouvement. Le tour réfléchi est donc plus fort, plus expressif. Mais le choix stylistique a vite entrainé la banalisation au point qu'il a fallu renforcer la réflexivité: se ipsum mouere (Flobert 1975: 387)
According to Flobert, in Virgil one can find the beginning of a metaphorical use: the personification of inanimate subjects, a phenomenon that is frequently used in late Latin:
Ce qu’on observe en latin tardif, surtout dans les textes techniques volontiers imagés, c'est une extension de plus en plus large du réfléchi pour décrire d'une façon dynamique des processus physiques touchant des êtres inanimés; la tendance est ancienne puisque l’on a déjà observé chez Virgile le mème phénomène dicté par un souci de personnification ou plus précisément d'animisation. (p. 389)
(19) frangitur inque sinus scindit sese unda, Virg. Aen.1, 161 breaks in-and curves divides itself wave
approx. ‘the wave divides itself and breaks in curves’ (Flobert: 388) The reflexive pronoun can also co-occur with verbs of change:
(20) convertit se in hirundinem (Plaut) turned himself into (a) swallow Moreover, se can accompany the forms in -r:
(21) repens (= serpens) torquebatur se, Vit, Anton, 9 (middle).
‘the snake twisted (itself) around’
(22) me nunc commoror, Plaut, Ps. 1131 (deponent)
‘as for me, I now refrain [myself]’
In brief, in Flobert’s words, the reflexive pronoun stresses the important role the referent of the subject has in the accomplishment of the activity (p. 388)
In Vulgar Latin, when it co-occurs with transitive verbs, the impersonal value also develops the possibility of expressing a similar value to that of a passive, because the object is subjectivized (see 23):
(23) litterae se scribunt letters REFL.3rd write:3rd.PL
‘one writes letters’–> ‘letters are written’.
The reflexive pronoun frequently occurs with verbs of movement:
(24) se ducere
lit. ‘to carry oneself [away]’, i.e. ‘to leave’
(25) se mouit ex urbe
REFL/3rd moved out-of town
If it serves as an indirect object, the reflexive pronoun may be found as an expression of empathy (the so-called emphatic dative):
(26) et sic, quia sera erat, gustavimus nobis (Per. Aeth., 4,8)
and so, since evening was-it we-had-dinner to us
‘and since it was evening, we had dinner’.
4. ROMANCE REFLEXIVE
If considered in syntactic terms, the Romance reflexive embraces a variety of structures in which two Roles are co-referential:
a. objective (transitive reflexive): Agent = Patient. Compare (27) and (28):
(27) Lat. cingor lavor
‘I girdle [myself]’ ‘I wash myself’
Romance:
(28) Fr je me lave versus je lave mon enfant
I myself wash. I wash my child
‘I wash’ ‘I wash my child’
Sp. me lavo me wash.
‘I wash‘
b. middle: Experiencer = Inner cause of feelings, states. Compare (61) and (62):
(29) Lat. Corinthia, quibus delectatur (Ciceron) Chorintinian which pleases
“Chorintian potery which pleases”
and Romance:
(30) Fr je me réjouis, Sp. me alegro, It. si rallegra, Rom. mă bucur
‘I rejoice/ I am glad‘.
The fact that all the verbs of feeling have a reflexive form reveals the fact that in Romance culture the feelings are conceived as rather dynamic processes than as [inner] states. The inner cause may be different from the real cause of the feeling (see 32), but the Inner cause and the Location of feeling are identical. The dynamicity of the feeling presentation in Romance is somehow lost in the English translation, which presents the feeling as a state:
(32) Rom. m-am bucurat de succesul tău
‘I was glad at your success‘.
In Romanian, middle reflexive may be found also with the verb of cognition such as ‘to think‘:
(33) Rom. se gândeşte mereu la tine REFL:3rd.ACC think-he always at you
‘he is thinking of you continuously’
c. reciprocal: Agent1= Patient1 and Agent2 = Patient2
The reciprocal reflexive has become rather productive in Romance languages:
(34) Fr. il se bat avec tout le monde he REFl:3rd/ACC ights with everybody
‘he fights with everybody’
Sp. se pelean It. si sono salutati
‘they fight with one another’ ‘they greeted each other’
Rom. se privesc
‘they look at each other’
Romanian has also a dative reflexive construction with verbs or expressions of knowing or feeling:
(35) Rom. Petru îşi aminteşte de tine Peter to himself remembers of you
‘Peter remembers you’
d. inchoative reflexive. As mentioned before, the inchoative value (i.e. the expression of change centered on the beginning of the new state) could be expressed in Latin either by a suffix: -esc (e.g. dulcescat ‘becomes sweet’) or by a form in -r (e.g. ignis exstinxitur ‘the fire extinguished [by itself]’). In Romance, the suffix -esc has lost its
inchoative value: cf. the present of French such as Fr. fleurir ‘to bloom’: il fleurit ‘it blooms’ – ils fleurissent ‘they bloom’, with the syffix , jouir ‘to rejoice’: je me réjouis
‘I rejoice’ – nous nous rejouissons ‘we rejoice’; or Rom. a munci ‘to work’: [eu]
muncesc ‘I work/am working’, with the suffix -esc, but noi muncim ‘we work/are working’, without the suffix. By turn, the reflexive construction has become the most common form of expressing a change:
(i) change in quality:
α) physical change:
(36) Fr. se rechauffer s’enrichir de la sueur du peuple
‘to get warm’ ‘to get rich thanks to the people's sweat’
(37) Sp. la nieve se derrite al sol (38) It. incalorirsi
‘the snow melts in the sun’ ‘to get warm’
β) Psychological verbs: Experiencer = Inner Cause of change:
(38) Sp. Juan se enfada de nada (39) Pg. emoborrastarse “John gets angry with no cause” “to get bored”
(40) It. offendersi (41) Rom. a se plictisi ‘to get offended’ ’to get bored’
(ii) change in position:
(42) Fr. se coucher se plier
‘to go to bed’ ‘to bent [oneself]’
(43) Sp. acostarse
‘to [make oneself to] lie down, to go to bed’
(44) Rom. a se apleca a se scula (45) It. muoversi
‘to bent’ ‘to get up’ ‘to move [oneself]’
e. impersonal: In the cases of impersonal and passive reflexives, the co-referentiality of two arguments is less obvious. Compare Vulg. Latin (23) and Romance (46):
(46) a. Sp. se sale hoy b. Rom. se pleacă azi
REFL:3rd leaves today approx. ‘the departure is today’
(47) It. Quando si è in ballo, bisogna ballare ‘when one is at the ball, one has to dance’
In (48) a so-called agreement mistake – which occurs frequently in spoken registers, se is treated as the subject and consequently the verb takes the third person singular even if the direct object is in the plural:
(48) Sp. se vende esas casas REFL.3rd sells these houses
‘these houses are for sale’,
See also spoken – substandard – Romanian:6
(49) se dă/ bagă/ măsline la Unic REFL:3rd gives/ enters olives at Unic
‘olives are given/ are entered in/ the shop [= sold] at the [supermarket] Unic’.
f. reflexive passive:
(50) Sp. las camisias de seda no se lavan facilmente the shirts of silk not REFL wash easily
‘the silk shirts do not wash easily’
(51) Rom. cămăşile de mătase nu se spală uşor shirts-the of silk not REFL wash easily
‘the silk shirts do not wash easily’
(52) It. in casa si accedono i lumi in house REFL light upthe lights
‘in the house the lights light up’
French makes a difference between intransitive and transitive constructions with a topicalized object. In (53) the whole utterance is a habitual expression and consequently les jambons ‘the hams’ are the topic, the controlling quality of usual necessity expressed by the verbal phrase:
(53) Fr. les jambons se pendent dans le grenier the hams REFL:3rd in the attic
‘hams have to be hung in the attic’
In (54) the whole utterance is rhematic:
(54) il se vend beaucoup d' articles anglais à Paris it REFL sells many items English in Paris
‘many English items are sold in Paris’
Although the syntactic approach is useful for desambiguating purposes, especially in contrastive grammars, for translation purposes, it has less explanatory power than the discourse hypothesis when it comes to account for the fact that the same construction can have so many functions.
6 Utterance (49) was often heard during the communist era when some goods were so rarely seen in the shops that when they were available for sale people considered them as a kind of gift – in spite of the fact that they have to be paid for.
The variety of interpretations of Spanish reflexives has been explored by Maldonado (1996) in interesting ways that support our hypothesis of a event- centered reflexive. On the one hand SE may serves to focus on the characteristics of the process and background the participant who has the initiative; on the other hand, it favors the interpretation of the event as an unexpected happening. Let us examine some of Maldonado's explanations.
a. Focussing on the process and backgrounding the initiative source
(i) For example, constructions expressing a change in body posture (see 55) focus on the change of state that is undergone and do not focus on the initiator or on the input.
Maldonado calls this tendency ‘terminal prominence’. This concept is important because it is one of the features of SE that applies to other constructions.
(55) Sp. Maria se arrodilló Maria REFL kneeled-down
‘Maria kneeled down’
With intransitive verbs SE also focusses on the change in state: Compare (60) and (61):
(60) Cuando llegué a casa, Juan ya se había ido.
When arrived-I at home Juan alreadyREFL had left
‘When I arrived home, Juan had already left’
(61) Juan /*se/ va al cine todos los días (habitual) Juan *REFL goes at-the movies all:Pl the days
‘Juan goes to the movies every day’
In (60), the speaker refers to the act of Juan going to the movies every day as a whole, thus IRSE would be inappropriate. In (61), we can only use IRSE because we focus on the moment of departure.7
(ii) The impersonal passive focusses on the change of state, whereas the initiative source is left in the background:
(62) se habla e spañol REFL speaks Spanish
‘we speak Spanish [is spoken]’
(63) la maleta se perdió en el aeropuerto the suitcase REFL losed in the airport
‘the suitcase got lost in the airport’
7 See also SUBIR and DORMIR: The more energy is involved in a verb, the more dynamic it becomes. Maldonado's hierarchy is: DORMIR > IR > SUBIR
The motivation for not specifying the doer varies: the doer may be conceptualized as non-specified external entity/force either because it is irrelevant (see 62) or the speaker wants to play down the responsibility of the doer (63).8
Maldonando points to the fact that the exact meaning of SE brings to the construction depends greatly on the semantics of the verb involved. SE is a polysemic element and not a meaningless detransitivizer as generative garmmars claim.
b. Cognitive/Emotion Middle
Maldonaldo distinguishes tow main groups:
(i) transitive verbs with Experiencers participant. In these cases SE decreases the transitivity of the verb: Usually humans are seen as being in control of their mental activities – to depart from this we can use SE to show a lower degree of participation:
(ii) The second group are transitive verbs of patient-like participants (feelings). SE increases the level of participation.
(64) me/le alegra verte (me V + Subject(S: verte))
and me/him pleases to see you
‘Seeing you pleases me’
(65) se alegra verte (reflexive) REFL.3rd enjoyes.he to see.you
‘He is glad to see you’
c. Inherent SE. These verbs have only a reflexive form (no active counterpart):
verbs which express a high degree of affective participation (Cf. Latin deponents):
Lat. queror ‘complain’. They differ from absolute verbs (such as llorar, suspirar) in that involve more participation or control on the part of the participant. Verbs such as llorar do not combine well with adverbs expressing active participation of their subject referent:
(66) *?ella lloraba agresivamente.
‘she was crying aggressively’
Inherent middle do accept such adverbs:
(67) se quejaba agresivamente
‘she was complaining aggressively’.9
8 Maldonado claims that this use of SE has become quite productive in Hispanic countries. He mentions a comedian that included in part of his routine to say: “It’s SE's fault!” (precisely because he wanted to refer to the way one does not want to be hold responsible for his wrong doings).
9 Maldonado claims that these verbs involve ‘attitude’: e.g. pavonearse (‘to strut’) and jactarse
‘to brag’ differ from caminar and hablar in the sense that they are a way of walking or talking with a specific attitude behind it.
5. REFLEXIVE PASSIVE VERSUS PLAIN PASSIVE
The hypothesis that the voice is the expression of the intersection between two parameters, the Role structure and the cognitive-discursive hierarchy, can explain the difference between th Romance plain passive and the reflexive passive.
(a) The active and the plain passive are expressions of a part-centered perspective, but differ as to the part characterized by the highest discursive prominence.
(i) The active voice, as the neutral term of the diathesis gives the highest discourse prominence to its first argument; consequently, its second argument may be deleted (e.g. dans ce restaurant on mange bien; on a bien mangé lit. ‘in this restaurant one can eat well, we have eaten [very] well’).
(ii) The plain passive is undergoer-centered. It explains the fact that the first argument can be deleted, and occurs only if it is focalized. (v. Desclès et al.
1985).
(b) The reflexive is, as its Latin counterpart (i.e. the forms in -r), event-centered (be it an activity or a state), which explains the fact that reflexive constructions may function as an “impersonal” (i.e. both the first and the second arguments of the verbs are omitted).
(68) esp. aquí se come bien; se sale hóy rom. aici se mănâncă bine; se pleacă azi
here REFL eats well; REFL leaves today
‘here one eats well’ ‘we are leaving today’
This hypothesis can also explain the following syntactic phenomena:
(a) the incompatibility of the reflexive – even passive – with an agentive by + NP. Limited to certain categories in Old Portuguese, Old Spanish Italian and Romanian, the agentive complement has become unacceptable.
(b) its preference for expressing telic (and not categorial) predicates (v. Moortgat 1990). Even when functioning as a qualitative (réflechi facilitant), its gives equal prominence to both the subject and the verb: e.g. le vin rouge se boit chambré). The quality entails the type of activity. Compare:
Spanish:
(69) En el palacio de Santa Cruz se firmó In the palace of Santa Cruz REFL:3rd signed un convenio [...] entre España y Italia
a treaty [...] between Spain and Italy
‘In the Santa Cruz Palace a treaty was signed between Spain and Italy’
Firmaron el Ministro español de Asuntos Signed:3.pl the Minister Spanish of Affairs Exteriores y el subsecretario italiano.
External and the subsecretary Italian
‘The Spanish Minister of Foreign Affairs and the Italian subsecretary of state signed [it]’
and
(70) La paz fue firmada por los embajadores de los dos paises peace was signed by the ambassadors of the two countries.
‘The peace [treaty] was signed by the ambassadors of both countries’
Reflexive is less likely in this context:
(71) ?Se firmó la paz por los embajadores de los dos paises.
?REFL:3 signed the peace by the ambassadors of the two countries.
Italian:
(72) Con un caretto se portavano le reti With a chariot REFL took the nets sulla spiaggia, si stendevano ...
to-the beach, REFL spread...
‘With a little chariot the nets were taken to the beach, they were spread...’
Romanian:
(73) În casă se face curăţenie mare In house REFL:3rd does cleaning big
‘big cleaning job is done in the house’
The passive constructions is less acceptable for expressing rhematic utterances:
(74) în casă este făcută curăţenie mare in house is done cleaning big
‘in the house big gleaning job is done’.
When the verb belongs to one of the following categories: verbs of saying, perlocutionary verbs (“to order”), verbs of ascertaining (“to ascertain”, “to notice”):
(75) Sp. se había comunicado [...] a los oficiales de infanteria que las primeras tropas de relevo ...
‘they were informed the officials of the infantry [division] that the first relief forces...‘
(76) Rom. li s- a ordonat să plece
to them REFL.3rd has ordered that leave. SUBJ
‘they were ordered to leave’
(77) Pg. Constatou- se que viera em vão acertained REFL.3rd that came.he in vain
‘it has been ascertained that he came in vain’
but
(78) Fue constatado por todos que viera em vão
‘everybody realized that he came in vain’
6. WHY HAS THE AGENT TO DISAPPEAR IN ORDER TO HAVE A NON-ACTIVE CONSTRUCTION?
A rich literature has underscored the fact that the reflexive construction is usually compatible with agentless sentences (cf. French ‘l’agent doit s'évader, s’évanouir, s’évincer’). Largely accepted as a given constraint, the necessity for the Agent to disappear from the focus of the speaker's consciousness raises some hot questions that cannot be answered by considering only syntactic an/or semantic levels.
Why has the Agent (i.e. the initiator/the most dynamic participant/ the first argument of the verb) has to disappear in order to let other arguments bear the highest discursive prominence, to be the center/the figure of the linguistic reconstruction of the event?
Why is the feature [Person] responsible for the choice of one of the passives? In our opinion, – as we have already emphasized – the linguistic reconstruction of the events follows a functional model that links grammar (syntax and/or morphology) to the inherent and/or contextual features of nouns and verbs, to the discursive organization as well as to the mentality of the speakers as human beings in general and as representatives of the triplet ego/hic/nunc in particular. These relations may be represented roughly as follows:
Scheme 1:
inherent: +Person features
contextual: +Dynamic/first argument
disc: +Topic
syntax: +Subject
word order: initial-sentence position
A similar idea can be also found in Traugott and Pratt (1980: 283):
...in English and many other languages, the most unemphatic form of language and the one with the least assumptions makes the following correlations in a sentence:
Theme Verb X
Agent Verb X
DefNP (shared) Verb (Indef)NP (unshared)
Subject Verb X”
In fact this functional model can account for the so-called argument hierarchy, what it is to be considered as the first or as the second argument of the verb. In brief, as soon as the Agent is present, it has the higher chances to become the figure [the center of the discourse]. An interesting proof of this relation is provided by the way in which grammars react when this relation is broken at least in one of its points. We have chosen to illustrate this point by considering some example from French, because it is not a pro-drop language (the preverbal pronominal subject has become obligatory) and consequently needs more markers to show that this relations are not observed:
Compare:
(a) When the subject represents to the topic and refers to a Person who does not play the role of the most dynamic participant, the plain passive is chosen:
(79) Paul a été éliminé de la course.
‘Paul was eliminated from the competition’
(b) When the referent of the topic is a Non-Person and a non-dynamic participant and co-occurs with a transitive verb, there are two possibilities.
(i) If the object is topicalized, usually if the co-enunciator repeats the noun in question as a signal of accepting the topic of the discourse proposed by the enunciator, then an active construction with a pronominal trace of the direct object is preferred:
(80) le gateau, je ne sais pas qui l'a mangé [Patient/Objective]
the cake, I don't know who it has eaten
‘the cake, I don't know who ate it’
(ii) One can choose the passive construction instead, usually if the whole sentence is rhematic:
(81) la terre était parsemée de fleurs [Locative]
the ground was sprinkled with flowers
‘the ground was sprinkled with flowers’
(c) With an intransitive verb, when the first argument is a person, conceived as dynamic participant, but it is not the topic (usually in a rhematic sentence), an empty subject occurs (il) and the verb agrees with this grammatical subject:
(82) il arrive des invités
it:masc/sg arrives INDEF ART guests
‘Guests are arriving’
(d) The same construction with an empty subject (il) is used when the unique argument of the intransitive verb is a non-Person/non-dynamic/non-topic:
(83) il arrive des trains
it:MASC.SG arrives INDEF ART trains
‘trains are arriving’
(e) In a transitive construction, when the direct object is not the topic, the reflexive pronoun -se co-occurs with il – the empty subject. In brief, the direct object is: non- Person/non-dynamic/non-topic:
(84) il se vend beaucoup d' articles italiens it:MASC.SG REFL:3rd sell many items Italian en Californie.
in California.
‘in California, many Italian items are sold ’.
CONCLUSIONS
Our hypothesis has the advantage of accounting for the following phenomena:
(a) The forms in -r may ignore the Agent [cause/instigator] altogether.
(b) In the category of verba affectuum (verbs of feeling), the subject corresponds to an Experiencer rather than to an Agent.
(c) As for the reflexive interpretation of the forms in -r (with verbs of movement or verbs expressing activities of body care, such as lavari, tergeri, etc.), the subject refers to more than one Role (Agent, Patient, Beneficiary, moved object, etc.).
But a change in orientation when reconstructing mentally and linguistically the event (or situation) cannot account on its own for the differences between the Latin category of voice and the Romance developments, in which at least three voices (active, passive and reflexive) if not four (active, passive, reflexive, factitive) came into existence. In our hypothesis, one has to consider also the changes that occurred in the encoding of various cognitive categories, which are highly culture-dependent.
Changes in cognitive categories have an important impact upon the process of reconfiguring the semantics structure of noun classes. In the case of the evolution of the category of voice, one has to consider the importance given to an inherent semantic feature such as [the Capacity of being a doer or not] in Latin, and to [Person]
rather than to [Agent] in Romance.10
REFERENCES
Aristotle, 1993, Aristotle’s Metaphysics. Books Γ, ∆, and Ε, Translated with Notes by Christopher Kirwan, 2nd edition, Oxford, Clarendon Press.
Bennett, C. E., 1910- 1914, Syntax of Early Latin. 1. The Verb, 2. The Cases, Boston, Allyn & Bacon.
Cowgill, W,, 1982, “On the Prehistory of Celtic Passive and Deponent Inflection” Ériu 33, 73–111.
10 See Aristotle for the definition of such a seme and Manoliu 2005; for the relation between topic and semes such as [+Person] and [+Dynamic] in Romance languages, see Manoliu, 1987.
Croft, W., H. B.-Z. Shyldkrot, S. Kemmer, 1987, “Diachronic Semantic Processes in the Middle Voice” in G. Ramat, A., O. Carruba, G. Bernini (eds.), Papers from the 7th International Conference on Historical Linguistics, Amsterdam/Philadelphia, Benjamins, 179–191.
Comrie, B., 1988, Passive and Voice, in M. Shibatani (ed.), Passive and Voice, Amsterdam/ Philadelphia, Benjamin, 9–23.
Ernout, A., F. Thomas, 19531, 1993, Syntaxe latine, 2e Edition, 8e tirage, Paris, Klincksieck.
Everaert, M., 1986, The Syntax of Reflexivization, Dordrecht, Foris.
Flobert, D., 1975, Les verbes déponents latins des origines à Charlemagne [Publications de la Sorbonne, Série NS, Recherches, 17), Paris, Belles Lettres.
Grandgent, C. H., 1962, An Introduction to Vulgar Latin, New York, Hafner.
Guillaume, G., 1971–73, Leçons de linguistique de Gustave Guillaume, publiées par Roch Valin, 1948–1949. Psychosystématique du langage. Principes, méthodes et applications, 1–2, 3, Québec/
Paris, Presses de l'Université Laval.
Jaszczolt, K. M., 2002, Semantics and Pragmatics. Meaning in Language and Discourse, London, Pearson Education/Longman.
Joffre, M.-D., 1995, Le verbe latin: voix et diathèse, Louvain/Paris, Peeters.
Kemmer, S., 1993, The Middle Voice [Typological Studies in Language, 23], Amsterdam, Benjamins.
Langacker, R. W. & P. Munro, 1975, “Passives and their meaning”, Language 51, 3-4, 784–830.
Langacker, Ronald, 1982, “Space Grammar, Analysability, and the English Passive”, Language 58, 22–79.
Langacker, R. W., 1987, Foundations of Cognitive Grammar. 1. Theoretical Prerequisites, Stanford, California, Stanford University Press.
Levinson, Stephen C., 1983, Pragmatics, Cambridge, London, New York, New Rochelle, Melborune, Sydney, Cambridge University Press.
Levinson, Stephen C., 2000, Presumptive meanings: The Theory of Generalized Conversational Implicatures, Cambridge, Mass., The M.I.T. Press.
Luraghi, Silvia, 1995, “Prototypicality and agenthood in Indo-European” in H. Andersen (ed.), Historical Linguistics 1993. Selected papers from the 11th International Conference on Historical Linguistics, Los Angeles, 16-20 August 1993, Amsterdam/ Philadelphia, Benjamins, 259–268.
Maldonado, R., 1993, “Dynamic construals in Spanish”, Studi italiani di Linguistica teorica e applicata, 22.3, 532–566.
Manoliu-Manea, M., 1987, “The Myth of the Agent: Roles and Communicative Dynamism in Romance”
in B. Cazelles and R. Girard (eds.), Alphonse Juilland: D'une passion l'autre, [Stanford French Studies], Stanford, Anma Libri, 261–275.
Manoliu-Manea, M., 1994, Discourse and Pragmatic Constraints on Grammatical Choices. A Grammar of Surprises [North Holland Linguistics Series 57], Amsterdam, Elsevier.
Manoliu, M. M., 2006, “Cognitive categories and Noun Classification. The Animacy Fallacy:
Redefining Neuter Gender” in Papers from the 4th Hawaii International Conference on Arts and Humanities, January 11–14, 2006, CD.
Meillet, A., 1921 (1) - 1937 (2), Linguistique historique et linguistique générale. Paris, Champion.
Meillet, A., J. Vendryes, 1960, Traité de grammaire comparée des langues classiques, 3e éd. rev. et comp. par J. Bedryes, Paris, Champion.
Milner, J. C., 1978, “Le système du réfléchi latin”, C. Touratier (éd.), Langages 50. Linguistique et latin, 73–85.
Perlmutter, D. et P. Postal, 1977, “Toward a Universal Characterization of Passivization” in Papers from the Annual Meeting of the Berkeley Linguistics Society, 394–417.
Sells, P., 2001, “Form and Function in the Typology of Grammatical Voice Systems”, in Geraldine Legendre, Jane Grimshaw, and Sten Vikner (eds), Optimality-theoretic syntax, Cambridge, Mass, MIT Press, 355–391.
Shlain, Leonard, 1999 (19981), The Alphabet versus the Godess. The Conflict between Word and Image, London/ New York, Penguib/Compass.
Touratier, Christian, 1984, “Il y a un passif en latin; mais de quoi s'agit-il?” in Le passif. Travaux 2 [Publications de l'Université de Provence] Bresson, Daniel (ed.), Laffite, Aix-en-Provence/
Marseille, 75–92.
Weiner, W., W. Labov, 1983, “Constraints on the agentless passive”, Journal of Linguistics, 19, 29–58.
Received March 2006
