Professional Documents
Culture Documents
M I C H EL A C EN N A M O
4.1 Introduction
This chapter investigates the morphosyntax of passive and impersonal reflexives in
contemporary Italian dialects and in some 11th–15th-century northern (Venetian,
Lombard), central (Florentine), southern (Neapolitan), and Sardinian (Logudorese)
vernaculars, considering the distribution of morphosyntactic, semantic, and prag-
matic characteristics which set several northern varieties apart from central and
southern dialects and from Sardinian.
It is shown that variational and diachronic data from early Italo-Romance throw
light on the diachronic processes leading to the rise of impersonal reflexives in
Romance, contributing to a better understanding of some highly debated issues of
Italian morphosyntax, such as the nature and function(s) of the reflexive morpheme
si ‘self-’, and the function of the clitic ci ‘us; there’ in the impersonal of reflexives.
The discussion is organized as follows. Section 4.2 illustrates the use of the reflexive
morpheme as a voice marker in Italian and some current theoretical assumptions on
its (varying) distribution and functions. Sections 4.3 and 4.4 describe the range of
variation in the use of the reflexive morpheme si/se in passive and impersonal
patterns in the Italian dialects and the early vernaculars, respectively. Section 4.5
* To Mair, a rigorous and critical mind, a ‘native speaker’ of early northern Italian vernaculars, and a
very generous and inspiring scholar and friend.
I wish to thank Paola Benincà, Adam Ledgeway, and Nigel Vincent for commenting on an earlier draft
of this chapter. I am also grateful to Diego Pescarini and Davide Ricca for discussions on Paduan and
Torinese, respectively, and to Ignazio Putzu and Maurizio Virdis for their help with old Logudorese. The
usual disclaimers apply.
Comp. by: PG2689 Stage : Proof ChapterID: 0002079044 Date:9/12/13 Time:17:45:34
Filepath:d:/womat-filecopy/0002079044.3D72
72 Michela Cennamo
builds on the insights gained from the analysis of the synchronic and diachronic
variation, and addresses the theoretical debate concerning the status of the reflexive
morpheme in impersonal patterns and the existence of either two types of si (a
passive and an impersonal one) or of only one si with different uses.
74 Michela Cennamo
In the passive pattern (5a) the past participle is in the masculine plural, agreeing
with an underlying unexpressed O—reflecting the more general rule whereby in the
impersonal pattern the nominal part of the predicate is in the plural (Salvi 1988: 98;
Cennamo 1993a: 36; 2010)—and the agent optionally surfaces as a prepositional
phrase introduced by the preposition da ‘by’. In the impersonal (active) construction
(5b), by contrast, the past participle occurs in the non-agreeing msg form.
This pattern of agreement is shared with divalent verbs with an unexpressed/
omissible object (e.g. pagare, ‘pay’, bere ‘drink’, leggere ‘read’), divalent verbs with an
‘oblique’ O (e.g. discutere di ‘talk about’, pensare a ‘think about’), and unergative
verbs (e.g. camminare ‘walk’, passeggiare ‘stroll’).
On the other hand, unaccusatives, equative patterns, and impersonal passives show
split agreement in compound tenses, with the finite verb displaying 3sg agreement and
the nominal part of the predicate mpl (or more rarely fpl) agreement:
(6) si è contenti/-e/insegnanti/ascoltati/partitI all’ alba (It.)
self= is happy.mpl/fpl/teachers/listened.to.mpl/left.mpl at.the dawn
‘one is/they/we are happy/teachers/ listened to/left at dawn’
Si can also mark the taking place of an event with no implied participant, in
conjunction with the verb fare ‘do’, e.g. far(si) tardi ‘become late’, with the past
participle occurring in the default msg in compound tenses (see e.g. Cennamo 1995):
(7) si fa tardi / si è fatto tardi (It.)
self= makes late / self= is made.msg late
‘it is getting late/it got late’
76 Michela Cennamo
between passive and impersonal si (}4.3.1), (ii) their grammatical domains (}4.3.2),
(iii) the nature of the subject (}4.3.3), (iv) tense/aspect restrictions (}4.3.4), (v) the
cooccurrence of the impersonal reflexive with a pronominal object (}4.3.5), and (vi)
the interpretation of the reflexive morpheme (}4.3.6; see also Cennamo 1997: 161).
78 Michela Cennamo
In some dialects (Torinese, Cairese, Genoese) (Battye 1990; Parry 1998a; 2005), the
patterns with and without verb agreement with the postverbal nominal (si V S/si
V N) alternate in passive and impersonal function. The pattern with agreement is
regarded, however, as less dialectal, reflecting Italian influence, as in (16a) from
Torinese (Cennamo 1997: 154):
(16) a. a s katu le turte (Tor.)
scl= self= buy the cakes
‘the cakes are bought’
b. a s kata le turte (Tor.)
scl= self= buys the cakes
‘one buys the cakes’
Interestingly, in some northern varieties (e.g.Cairese and generally in Piedmontese
and Ligurian) the morphological/syntactic impersonality of the pattern does not
entail its impersonal interpretation (Parry 1998a; 2005) (17):
R
(17) a. i s bøtu i pjat au so po t (Cairo Montenotte (SV), Liguria)
scl= self= put the plates at.the their place
‘the plates are put away’
b. u s bøt i pjat au so
scl.imprs= self= puts the plates at.the their
R
po t (Cairo Montenotte (SV), Liguria)
place
‘the plates are put away/one puts the plates away’
In some Piedmontese varieties (e.g. Borgomanero) impersonal si occurs preverb-
ally on a par with subject clitics (18a), whereas anticausative and passive si occur
postverbally (18b,c) (Manzini and Savoia 2005: ii.70–71):
(18) a. ki a z droma bei (Borgomanero (NO), Piedmont)
here scl= self= sleeps well
‘here one sleeps well’
b. i 'rumpusi I pjati (Borgomanero (NO), Piedmont)
scl= break=self The plates
‘the plates break’
c. lø i vøŋgusi sempri pasøndu (Borgomanero (NO), Piedmont)
they scl= see=self always going
‘they are always seen going . . . ’
Other northern dialects such as Bussoleno (Piedmontese) (19) and Friulan show
free variation of pre-/postverbal subjects, although the S si V/si V S order generally
conveys a pragmatic distinction, the distribution of given-new information
(Cennamo 1997: 157; 1998):
Comp. by: PG2689 Stage : Proof ChapterID: 0002079044 Date:9/12/13 Time:17:45:38
Filepath:d:/womat-filecopy/0002079044.3D79
80 Michela Cennamo
82 Michela Cennamo
1
The bibliographical details of the textual sources and their abbreviations cited in the examples below are
as follows: Bonvesin, Disputatio = Disputatio musce cum formica, in Contini (1937: 27–40); Bonvesin,
Miraculis = De quindecin miraculis que debent apparere ante diem iudicij, in Contini (1937: 41–44); Capitolare
= Princivalli and Ortalli (1993); Capitoli = ‘Capitoli della Compagnia di San Gilio’, in Schiaffini (1926: 34–54);
Cedola di Marco Granello = ‘Cedola di Marco Granello’, in Stussi (1965: 36–40); Cedola di Marco Michel =
‘Cedola di Marco Michel’, in Stussi (1965: 111–22); Cedola di Pangrati Barbo = ‘Cedola di Pangrati Barbo
(Prima)’, in Stussi (1965: 63–4); Cronica = ‘Cronica degli Imperadori’, in Ceruti (1878); Cronica fior. = ‘Cronica
Fiorentina’, in Schiaffini (1926: 82–150); CSNT = Il Condaghe di San Nicola di Trullas, Merci (1992); CSPS = Il
Condaghe di San Pietro di Silki, Delogu (1997); RSPS = Il Registro di S. Pietro di Sorres, Piras and Dessì (2003);
Dante, Inferno = Petrocchi (1966); Grisostomo = ‘Neminem laedi nisi a se ipso’, Forster (1880–83); LDT =
Libro de la Destructione de Troya, De Blasi (1986); Libro = ‘Libro degli ordinamenti della Compagnia di Santa
Maria del Carmine’, in Schiaffini (1926: 55–72); Milione = Il Milione di Marco Polo, Bertolucci Pizzorusso
(1975); Milione veneto = Marco Polo. Il Milione Veneto, Barbieri and Andreose (1999); Passione = ‘La Passione
e altre antiche scritture Lombarde’, Salvioni (1886); Rettorica = Brunetto Latini, La Rettorica, Maggini (1968);
Ricordi= ‘Ricordi di compere e cambi di terre in Val di Streda e dintorni’, Castellani (1982: 215–54); SRS = ‘Gli
Statuti della Repubblica sassarese’, Guarnerio (1892–4).
2
The corpus investigated consists of non-literary texts from the 11th to 15th centuries, from different
areas, namely Old Venetian, Old Lombard, Old Neapolitan, and Old Logudoreses Sardinian, partly
available at http://www.ovi.cnr.it/. The choice follows the approach by Vincent, Parry, and Hastings
(2004). For Old Neapolitan we have benefited also from data from Adam Ledgeway. Abbreviations are
as follows: This chapter uses the following abbreviations for texts cited: Cronica = A. Ceruti (ed.), ‘Cronica
degli Imperadori’, Archivio glottologico italiano 3 (1878): 177–243. CSNT = P. Merci (ed.), Il Condaghe di San
Nicola di Trullas (Sassari: Delfino, 1992). CSPS = I. Delogu (ed.), Il Condaghe di San Pietro di Silki: testo
logudorese inedito dei secoli XI–XIII (Sassari: Dessì, 1997). Grisostomo = W. Forster (ed.), Antica parafrasi
lombarda del “Neminem laedi nisi a se ipso” di S. Giovanni Grisostomo’, Archivio glottologico italiano 7
(1880–83): 1–120. LDT = N. De Blasi (ed.), Libro de la Destructione de Troya: volgarizzamento napoletano
trecentesco da Guido delle Colonne (Rome: Bulzoni, 1986). Milione = A. Barbieri and A. Andreose (eds),
Comp. by: PG2689 Stage : Proof ChapterID: 0002079044 Date:9/12/13 Time:17:45:40
Filepath:d:/womat-filecopy/0002079044.3D84
84 Michela Cennamo
highly varied picture as regards the use of the reflexive morpheme in passive and
impersonal function, and appears to confirm the existence of a correlation between
the spread of this strategy in impersonal function and the referential domain of si/se.
In our discussion we do not consider ‘impersonal’ se/si in fixed phrases with
avalent/monovalent verbs like old Venetian convenire ‘be advisable’, contenere ‘con-
tain, state’, deser ‘become’, and their equivalents in the other vernaculars, well
attested in the early varieties investigated, sometimes preceded (according to the
verb) by expletive el ‘it’ and its variants in Old Venetian, Old Lombard, and
Old Florentine, alternating with non-reflexive forms (e.g. OVnz. (el) se convien(e)
‘(it) self= is.advisable’ ~ conven ‘is.advisable’).
The S–se–V order usually favours a passive interpretation, whilst the se–V–S
sequence favours an impersonal reading. The distinction between passive and imper-
sonal reflexive, however, is not yet fully grammaticalized, as shown in (30), where
either order is possible, although the subject conveys given information:
(30) a. tanto que questi dr. se paga (OVnz., Cedola di Marco Granello,
such that these denarii self= pay.3
39.24–5)
‘so that this money is paid’
Marco Polo: Il Milione veneto (Venice: Marsilio, 1999). Panfilo = A. Tobler (ed.), ‘Il Panfilo in antico
veneziano’, Archivio glottologico italiano 10 (1886): 179-243. Passione = C. Salvioni (ed.), ‘La Passione e altre
antiche scritture lombarde’, Archivio glottologico italiano 9 (1886): 1-22. Testi Ven. = A. Stussi (ed.), Testi
veneziani del Duecento e del Trecento (Pisa: Nistri Lischi, 1965).
Comp. by: PG2689 Stage : Proof ChapterID: 0002079044 Date:9/12/13 Time:17:45:40
Filepath:d:/womat-filecopy/0002079044.3D85
b. avanti che se paga sta dota (OVnz., Cedola di Marco Michel, 112.22)
before that self= pay.3 this dowry
‘before this dowry is paid/one pays/they pay this dowry’
The reflexive in impersonal function is confined to bivalent verbs with an implicit
latent object, sometimes preceded by expletive el ‘it’ (and its variants elo/eli) (31):
(31) sì co’ elo se trova (OVnz., Cedola di Pangrati Barbo, 63.8)
so as scl.imprs= self= find.3
‘as one finds it’
In addition, passive/impersonal and impersonal se only occur in simple tenses
and in imperfective verb forms, occasionally with the expletive el and a postverbal
subject (32):
(32) quando el se leze Henrico Primo (OVnz., Cronica, 226.18)
when it= self= read.3 Henry first
‘when one reads Henry the First’
In 13th-century texts impersonal se never occurs with monovalent verbs and
always refers to an indefinite human participant who does not include the speaker.
Thus se has either a generic or an indeterminate/existential reference, but never an
inclusive one. There are no examples of overt expression of the agent.
Impersonal se with monovalent verbs is attested from the mid- to late 14th
century and during the 15th century in imperfective tenses only, with non-
inclusive (33a) and inclusive indeterminate/existential reference (33b) (cf. also
Wehr 1995: 106):
(33) a. la calle che se va in Riolto Novo (OVnz., Capitolare 124, 58.18–19)
the street that self= go.3 to Riolto new
‘the path through which one goes/people go to Riolto Novo
b. torneremo ala provinzia [ . . . ], perché da altra via non
we.will.return to.the province since from other way not
se pò andar (OVnz., Milione veneto, 35.10)
self= can.3 go.inf
‘we will go back to the province [ . . . ] since we cannot go another way’
86 Michela Cennamo
In Grisostomo (Old Pavese) there are few occurrences of impersonal reflexive with
monovalent verbs in simple (imperfective) tenses only:
(38) a. ma per le citae se çeva pur peçorando (OPav., Grisostomo, 40.30)
but through the city self= went also worsening
‘but in the town things were getting worse’
b. onde se çoga soto a la roçça (OPav., Grisostomo, 15.22)
so.that self= plays under to the rock
‘so that one plays/they play under the rock’
88 Michela Cennamo
As for its interpretation, impersonal si can have both a generic and indeterminate/
existential (both exclusive and inclusive) reading. It may refer, in fact, to both
speaker and hearer/interlocutor, as shown by the alternation of si with the 1st person
plural (44):
(44) Demoli [ . . . ] a Iacopo Lanberti, ke gli si scontò
we.gave=them to Iacopo Lamberti that to.him= self= deducted
de’ denari ke Ci dee dare (OFlo., Ricordi, 224.13–14)
of.the denarii that us= must give.inf
‘he gave them (= the money) to Iacopo Lamberti, money that was deducted
from the denarii that he owes us’
90 Michela Cennamo
The expression of the agent is attested later, in 14th- to 15th-century texts (51a),
where passive/impersonal si also occurs with animate subjects (51b), occasionally in
perfective tenses (51c), albeit only with bivalent verbs, never with monovalent ones.
With the latter, impersonal si occurs mainly with unaccusatives like andare ‘go’ (51d),
always with a non-referential, exclusive interpretation:
(51) a. sas quales se deuen dare daue chalunque
the which self= must give.inf by whatever
persone (OLog., SRS XXVIII.8)
person
‘which must be given by everybody’
b. qui non si podiat hobligare sensa licentia
who not self= can force.inf without permission
desu p[re]ladu (OLog., RSPS (18r.) 82.16-17)
of.the priest
‘who one cannot force without his priest’s permission’
c. qui pius non si li siat dadu
who ever not self= to.him= be.sbjv given
termen (OLog., RSPS (64r.), 290.19-20)
limit
‘that one has never given him a temporal limit’
d. fina assa uia per issa quale se uaet
until to.the path through it which self= goes
ad OSilo (OLog., SRS XXXIII.7)
to Osilo
‘as far as the path by which one goes to Osilo’
92 Michela Cennamo
position, preceding all types of object clitic, unlike reflexive and passive si
(cf. Manzini and Savoia 2005: ii.19–62).
Diachronic data, on the other hand, appear to support the claim concerning the
existence of two different homophonous si’s: a reflexive ~ passive/impersonal si and
an impersonal si. In point of fact, whereas passive/impersonal si is attested in all the
early vernaculars investigated, impersonal si is hardly found in Old Logudorese
(where it is not attested in the early texts), and in both Old Logudorese and Old
Venetian it is found in later texts than passive/impersonal si. Indeed, synchronic and
diachronic variation support the findings in Cennamo (1993a; 1993b; 2000) whereby
passive and impersonal reflexives are two different structures, reflecting two different
diachronic paths which at some point merge. While the passive reflexive is a Late
Latin development, related to changes in the encoding of voice in the transition from
Latin to Romance, the impersonal function of the reflexive with one-argument verbs
is a Romance phenomenon, not equally attested in the Romance languages and in
the Italian dialects. More specifically, the rise of impersonal si in Italo-Romance may
be related to a stage where the reflexive pronoun acquires a non-anaphoric pronom-
inal value, whereby Latin se becomes equivalent to is ‘he’ (52)—a pattern that is well
attested in Italo-Romance, both diachronically and synchronically (Rohlfs 1968: }480,
458, n. 1; Manzini and Savoia 2005: ii.121–4)—also referring to 1st and 2nd person
participants in its reflexive/middle use in late texts, as illustrated in (53a,b), where se
is equivalent to nos ‘we’ and even me ‘I’ (Cennamo 1991: 17; 1993a: 81):
(52) ipsi [ . . . ] sibi crediderunt (Lat., Trad. Frising. 553; Cennamo 1991: 8)
they self.dat= believed
‘they trusted him’
(53) a. inter se singuli dissimiles
between themselves each different
invenimur (Lat., Min. Fel., 18.15; II AD)
we.find.pass
‘we find ourselves different from each other’
b. Ego [ . . . ] Adhalhardus [ . . . ] recogitans
I Adhalhardus thinking
se (Lat., Tardif: N: R 59; Cennamo 1993b: 58)
self=
‘I Adhalhardus, [ . . . ] thinking [ . . . ]’
Indeed, a possible starting point for the acquisition of the impersonal function of
the continuators of Latin se might have been the non-anaphoric pronominal function
and the gradual widening of the referential domain of si/se in the course of time,
whereby se/si comes to include the 1st and 2nd person participants, due to the 1pl
interpretation that the reflexive morpheme acquires in the course of time in its
Comp. by: PG2689 Stage : Proof ChapterID: 0002079044 Date:9/12/13 Time:17:45:44
Filepath:d:/womat-filecopy/0002079044.3D94
94 Michela Cennamo
impersonal function. This parallels a change that had already occurred in Latin in its
reflexive/middle use, whereby Latin se = nos and even me in late texts alongside the
non-reflexive, anaphoric pronominal use of se, whereby se = is, as illustrated in (52)
and (53).
As a result of the above-mentioned factors—(i) its use in non-anaphoric pronom-
inal function, (ii) the gradual widening of its referential function, and (ii), the gradual
grammaticalization of its ‘new’ referents, speaker and hearer, as in its 1pl interpret-
ation—se/si becomes an indefinite pronoun, and gradually ousts the homo ‘man’ type
in the early Italian vernaculars, which could only occur with a generic interpretation,
never an existential one (Egerland 2003; Manzini and Savoia 2005: ii.132). Thus the
impersonal use of the reflexive morpheme with one-argument verbs and various
types of impersonal pattern appears to be related to a series of changes where the
Person Hierarchy is one of the features involved (Cennamo 1993a: 81; 1993b: 60).
4.6.2 Impersonal ci
The analysis of the synchronic distribution of the 1pl reflexive clitic ci in impersonal
function in some Italian dialects points to the use of ci as an indefinite pronoun in
some varieties on a par with si. In some varieties ci is equivalent to si in reflexive,
anticausative, passive, and impersonal function. We can hypothesize a stage where se
and ci become functionally equivalent, competing in some areas, with ci replacing si
as a marker of agent defocusing/argument suppression. Thus in the ci–si–V patterns
as in ci si lava - ci= self= washes (‘one washes oneself ’), ci si pente – ci= self= repents
(‘one repents’), ci may be regared as an impersonal pronoun and si as the reflexive
pronoun, following the traditional description.
The alternative view, put forward in Cinque (1988a), Salvi (2008a; 2010b), and
Bentley (2006), regards ci as the 1pl pronoun and si as the impersonal pronoun.
According to Salvi (2008b; 2010a) the ci si lava pattern originates in Tuscan (noi) si
lava lit. ‘(we) self= washes’/ci si lava lit. ‘ci= self= washes’ = laviamo lit. ‘we.wash’/ci
laviamo lit. ‘us= we.wash’. It is attested in 19th-century northern writers such as
Rovani and Nievo (Salvi 2008a: 17; 2010b) and might have developed as a result of the
interference between the native language of these northern writers, where the
impersonal reflexive only had a generic/indeterminate, exclusive interpretation,
and the higher-prestige Tuscan variety on which the literary language is based
(Florentine), where the impersonal reflexive pattern could occur also with an inclu-
sive interpretation. Therefore northern writers/speakers chose the system of rules of
the higher-prestige variety that had a wider range of functions than their native
varieties. This might have led to the innovations introduced by northern writers/
speakers (one of which is the ci–si–V pattern), which then became consolidated in
literary and spoken Italian (Salvi 2010b: 7).
Comp. by: PG2689 Stage : Proof ChapterID: 0002079044 Date:9/12/13 Time:17:45:45
Filepath:d:/womat-filecopy/0002079044.3D95
4.7 Conclusions
The synchronic and diachronic data investigated show that the spread of the mor-
pheme si (and its variants) to impersonal function might be related to the gradual
extension of its referential domain and its degree of grammaticalization.
There appears to be a correlation between the syntactic contexts in which the
impersonal reflexive occurs and its generic or indeterminate/existential inclusive
interpretation. Dialects where the reflexive never acquires an indefinite/existential
inclusive interpretation show a limited use of this strategy in impersonal patterns.
More specifically, the data point to the existence of two different si’s, reflecting
diffferent diachronic paths converging into an apparently unitary pattern. While
the passive reflexive is a Late Latin development, related to changes in the encoding
of transitivity in the transition from Latin to Romance, the impersonal function of
the reflexive with monovalent verbs is a Romance phenomenon, not equally attested
in the Romance languages and in the Italian dialects, related to a stage where the
reflexive pronoun acquires a non-anaphoric pronominal value, whereby Latin se = is,
occurring also to refer to 1st and 2nd person participants (se = nos).
As for ci in the ci–si–V patterns in Standard Italian, it can be regarded as an
indefinite pronoun, having an indeterminate/existential interpretation, either inclu-
sive or non-inclusive, rather than the phonologically conditioned allomorph of si.