Professional Documents
Culture Documents
By Michela Cennamo
Department of Modern Philology, University of Naples
(Received 25 October 1997, revised 28 December 1998)
Abstract
This paper illustrates the relevance of the Unaccusative
Hypothesis for a well-known area of Late Latin/early
Romance syntax, the proliferation of pleonastic reflexives
with intransitive verbs denoting change of state/location,
states, verba dicendi and sentiendi. In particular, it is argued
that at some point in Late Latin the accusative and dative
reflexive pronouns (se/sibi) become markers of Split Intrans-
itivity, with se occurring with unergative/class SA verbs, and
sibi with unaccusative/class SO verbs. It is also shown that a
gradient approach to Unaccusativity/Split Intransitivity
accounts neatly for the data, allowing one to locate the
verbs/patterns under scrutiny along a Hierarchy of Un-
accusativity/Unergativity, resulting from the interplay of a
number of parameters, with Telicity and Control being most
relevant in defining the core of the categories.
1
Earlier versions of this paper were presented at the XIIth International Conference
on Historical Linguistics, in Manchester, 13 August to 18 August 1995, in the
Department of Greek and Latin of Manchester University in November 1995, and
at the IXth International Conference of Latin Linguistics, in Madrid, 14 April to 19
April 1997. I wish to thank the three audiences, in particular Jim Adams, Kate
Burridge, Robert Coleman, Christian Lehmann, John Payne, Mario Saltarelli and
Thomas Shannon. I have greatly benefited from lengthy and informative discussion,
at different stages, with Jim Adams, Phil Baldi, Oscar Collinge, Arturo De Vivo,
Harm Pinkster, Giovanni Polara, Valeria Viparelli, all of whom I wish to thank for
their comments on earlier versions of the paper. I am also grateful to Antonio
Nazzaro and Paola Santorelli for their help with the Christian Latin data, and to
Martin Maiden, Mair Parry, and Antonella Sorace with whom I discussed some
aspects of the work. Thanks are also due to two anonymous referees for their useful
comments, to Nigel Vincent for his constant encouragment, and to Keith Brown for
his helpful assistance. All shortcomings and misinterpretations are, of course, mine.
1. Introduction
The non-homogeneous nature of intransitives crosslinguistically is a
well-known phenomenon (already noticed by Sapir (1917) for some
Amerindian languages), referred to with a variety of labels, Split
Intransitivity, active-stative, active-inactive patterning, Unaccusa-
tivity/Unergativity, depending on the theoretical perspective
adopted (cf. Merlan 1985; Mithun 1991, among others, and refer-
ences quoted therein).
In several languages, in fact, the sole argument (S) of some
intransitive verbs shares semantic/morphological/syntactic proper-
ties with the Patient-like nuclear argument (O) of canonical transitive
verbs, whereas the sole argument of other intransitive verbs is
semantically/morphologically/syntactically akin to the Agent-like
nuclear argument (A) of canonical transitive verbs (with S, A, O
being syntactico-semantic primitives, referring to the core arguments
of a clause, coinciding, in their prototypical encoding, respectively,
with the notions of subject and object, in the languages where these
relations can be identified (Dixon 1979; 1994: 6±8; Foley & Van
Valin 1984; Andrews 1985; Van Valin 1990; 1993, among others) ).
In the languages where it holds, the distinction can be coded
either morphologically, as in Acehnese (an Austranesian language
of North Sumatra) ( (1) and (2) ), or it can be morphosyntactically
manifested, as in Italian ( (3) and (4) ).
In Acehnese the S argument of some intransitive verbs (e.g. jak
`go') (1a) takes the same cross-referencing pronominal proclitics on
the verb (geu for third person, loÃn for first person) as the A argument
of transitive verbs (1b), whereas the S argument of other intransitive
verbs (e.g. rheÈt `fall') (2a) shows optionality/absence of cross-
referencing pronominal enclitics on the verb, like the O argument
of transitive verbs (2b) (Durie 1987: 366±70).
(1) a. geu2-jak gopnyan
3-go (s)he
`(S)he goes'
2
The following abbreviations are used in the analytic glossing of the examples and
in the tables: A=accusative, AB=ablative, ACT=Actor, antic.=anticausative,
D=dative, F=feminine, G=genitive, GER=gerund, GERD=gerundive, I=infinitive,
IMP=impersonal, IMPER=imperative, IN=inchoative, inher. direct.=inherently
Actor Undergoer
!
Agent Effector Experiencer Locative Effector/Theme Theme Patient
2. Latin reflexives
We shall now illustrate the domain of Latin syntax which appears to
show the diachronic relevance of the Unaccusative Hypothesis,
namely `pleonastic reflexives'. The term is usually used in the
literature to label a variety of different uses of the reflexive pronouns
se/sibi, respectively the accusative/ablative and dative forms of the
reflexive pronoun (cf. also Cennamo 1991; 1995). Our analysis will
proceed from their canonical functions, some of which are semanti-
cally motivated (section 2.1), to the non-canonical and pleonastic
ones (sections 2.2, 2.3), showing, whenever possible, their inter-
action with the intermediate stages of their grammaticalization
process and the different areas of the grammar they belong to.
5
This notion, reflecting the degree of `primary responsibility' (Lakoff 1977) of a
core argument of the clause in the verbal process, results from the interplay of
different parameters, involving the inherent characteristics of the verbal argument(s)
(e.g., Animacy) and its/their thematic relation(s) (e.g., Agent, Patient, Theme) (cf.
Lehmann 1988; Comrie 1989: 59ff; Klaiman 1991a; 1991b: 110ff ).
6
In particular, Unaccusativity can be regarded as a `covert' type of active stative
patterning (Klaiman 1991b: 128) and Split Intransitivity as an overt one (cf. also
Cennamo 1995).
(12) nam mihi iam intus potione iuncea onerabo gulam (Plaut. Stich.
639)
in-fact I (D) now inside drink (AB) of-rush (AB) will-load
(1SG) gullet (A)
`in fact I will now load my gullet inside with a rush-drink'
one and the same area, one that seems to show patterns of active-like
typology also in the domain of verbal syntax, pairing the patterns of
active alignment of grammatical relations attested in Late/Medieval
Latin, discussed at length by Plank (1985: 287±93; 1995: 1193) (see
also La Fauci 1991; 1994: 21±49; Zamboni 1998 for a recent
discussion of the issue of (Late/Medieval) Latin active forms of
morphological organization surviving in Romance morphosyntax).
(17) a. ita nunc ignorans suo sibi servit patri (Plaut. Capt. 50)
so now unaware (N) his (D) RFL serves father (D)
`so now he is serving his father unaware'
b. suo sibi suco vivunt (sc.coclae) (id. 81)
their (AB) RFL juice (AB) live (3PL) (snails)
`they (=snails) subsist on their own juice'
c. me malum esse oportet . . . atque hunc telo suo sibi, malitia,
a foribus pellere (Plaut. Amph. 269)
I (A) bad (A) be (INF) is-necessary (IMP) and this (A)
weapon (AB) his (AB) RFL slyness (AB) from doors (AB)
drive-away (INF)
1964: 193), at times alternating, with the same meaning, with the
phrases a se/de se (literally `by oneself') in the Inscriptions (cf. Pirson
1901: 174±5; Konjetzny 1908: 194±5) (for a different interpretation
of these phrases, see DahleÂn 1964: 194±5):
(22) a. sed solus sibi immotus Atreus constat atque . . . deos terret
minantes (Sen. Thy. 703±4)
but alone (N) RFL still (N) Atreus (N) stands and gods (A)
threatens menacing (A)
`but Atreus stands alone, still, and threatens the menacing
gods'
(23) a. marito sibi merentissimo (CIL XII, 3453)
husband (D) RFL very-meritorious (D)
`to (her) husband himself highly meritorious'
b. uxori a se merentissimae (CIL XII, 194)
wife (D) by RFL very-meritorious (D)
`to (his) wife herself highly meritorious'
c. uxori de se meritae (CIL XII, 2OO)
wife (D) by RFL meritorious (D)
`to (his) wife herself meritorious'
By this time sibi can also occur (though more rarely) with [-An]
participants (24) under the same meaning (`(by) itself '), equalling
the phrase per se (25), which is however the form most typically used
(also attested under the meaning `only' (25d) during the 4th±5th
centuries A.D.), a usage which will become more common in later
centuries (cf. LoÈstedt 1911: 335; Svennung 1935: 316±8):
(24) nam scorpio sibi ipse pulcherrimum medicamentum est (Cels.5,
27, 5)
in-fact scorpio (N) RFL itself (N) excellent (N) drug (N) is
`in fact scorpio itself is an eccellent drug'
(25) a. Medicamentum est etiam vel plantaginis succus per se, vel
marrubii cum melle incoctus (id.3, 22, 35)
remedy (N) is also either plantain (G) juice (N) by RFL or
horehound (G) with honey (AB) cooked (PP.M)
`A further remedy is either plantain juice by itself, or a
horehound juice cooked with honey'
(27) a. contra naturam est, quae pedes dedit, ut per nos ambularemus,
oculos, ut per nos videremus (Sen. epist. 55, 1)
against nature (A) is that feet (A) gave (SG) so-that by
RFLwalked (1PL.SUBJ) eyes (A) so-that by we (A) saw
(1PL.SUBJ)
`it is against nature, that gave us legs for us to walk by
ourselves, eyes for us to see by ourselves'
b. dicens quod per semetipsa moverentur (sc.animalia) (Tert.
nat. 2, 3, 11; Corp. Christ. 1, 45)
saying (PRP.N) that by themselves (A) moved (3PL.SUBJ)
(animals)
`saying that they (=the animals) moved by themselves'
This `spontaneous' reading of sibi equalling per se (reported by
DahleÂn (1977: 15) as probably due to Greek influence) is already
attested in Itala (though apparently confined to adjectives denoting
change of state such as sibi mortuus (28a) and is quite common in 4th
century technical works (eg. Mulomedicina Chironis), and in Chris-
tian writers, with [+An] subjects (cf. Svennung 1935: 317; DahleÂn
1964: 116). It occurs in particular with intransitive verbs denoting
change of state (29) and location (30), with anticausatives (31) and
state verbs (32):
(28) a. ex carnibus eorum non manducabitis et sibi mortua eorum non
tangetis (Itala Lev. 11, 8, cod. Lugd.)
from meat (AB.PL) they (G) not will-eat (2PL) and RFL
dead (N.PL) they (G) not touch (2PL.SUBJ)
`do not eat their meat and do not touch their flesh if they
died naturally'
b. ut sibi mortuae (sc. haemorroidas) caderent (Soran. 2, 30,
83)
so-that RFL dead (N.PL) (haemorrhoids) fell (3PL.SUBJ)
`so that they (=haemorrhoids) fell spontaneously'
(33) et non post multum solet (sc. iumentum) sibi refrigerare (Chiron
414)
and not after long uses (donkey) RFL cool-off (INF)
`and after not too long it (=the donkey) uses to cool off by
itself'
Change of state verbs Change of location verbs State verbs Mental process verbs Speech act verbs Other activity verbs
(inher. direct.) Location Relation
sibi perire sibi vadere sibi vadere sibi manere sibi esse se poenitere se desperare se lamentare se vacare
sibi nasci sibi fugere sibi sedere se stupere se confidere se flere se pausare
sibi fieri sibi abire sibi stare se plorare se contremulare
sibi obdormiscere sibi discedere se periurare se obdormire
*se nasci se reclamare *sibi vacare
*sibi periurare
(anticausatives) (non-inher. direct.)
sibi combibere/se combibere sibi ambulare se insidere se esse
(sibi laxare/se laxare) sibi vagari se desidere
(sibi sanari/se sanare) sibi peregrinare se adhaerere
situation: situation:
telic, dynamic, concrete, telic (inher. direct.) atelic, static, telic/atelic, dynamic/static, atelic, dynamic, atelic, dynamic,
atelic (non-inher. direct.) concrete /abstract abstract concrete concrete
dynamic, concrete
subject: Subject:
UNDG/Patient ACT/Theme UNDG/Theme ACT/Experiencer ACT/Agent ACT/Agent
-Control (+Control-antic.) +Control +Control -Control +Control +Control
that canonical uses of fully pleonastic sibi occur with verbs having
telic aspect and an Undergoer subject, denoting a dynamic, concrete
situation. Canonical uses of fully pleonastic se, on the other hand,
are attested with inherently atelic verbs having an Actor subject,
denoting concrete, dynamic situations.
The interplay of the parameters of Aspect, Control, Dynamic/
Static, Concrete/Abstract, enables us to locate the patterns under
scrutiny along a continuum of Unaccusativity/ Unergativity, illus-
trated in table 2. High unaccusatives are characterized by the
presence of all parameters but Control, as with change of state
verbs, having the features telic, dynamic, concrete, -Control. Low
Unaccusativity on the other hand is characterized by the absence of
the above-mentioned parameters, namely by the features static,
abstract, -Control, atelic (sibi esse), with intermediate points char-
acterizable as concrete, static, -Control situations, as in sibi manere,
or dynamic, telic/atelic, concrete, +Control situations (as in sibi
vadere, sibi ambulare). High unergatives, on the other hand, denote
atelic, dynamic, concrete situations with an Actor subject. More
peripheral unergatives denote atelic, static, abstract situations with
an ACTOR/Experiencer subject, as in mental process verbs, such as
se desperare. Intermediate points are instantiated by telic, dynamic,
abstract situations, as in se poenitere/se stupere.
Fluctuations in the use of se/sibi appear to occur at the periphery
of the categories of Unergativity/Unaccusativity, in particular with
state verbs denoting location/relation, that is with atelic, concrete/
abstract, -Control, static situations (e.g., sibi manere, sibi stare, sibi
esse vs se residere, se desidere, se inhaerere, se esse), so that we either
have alternations of the two forms (53a±b) (both of the 6th century
A.D.), or we find some intransitive verbs taking se, with others
taking sibi, all other features being equal, as in (43a±d) vs (37a±e)
above, a fact which might reflect the different historical paths
pointed out above (cf. sections 2.2±2.3).
(53) a. = ( 37b) ostendit unamquamque in Trinitate personam . . .
ipsam sibi esse quod habere naturaliter predicatur (Fulg.
Rusp. epist. 14, 16; Corp. Christ. 91, 403)
through lexical classes, in that the texts show their occurrence with
most of the afore-mentioned classes by the 3rd±4th centuries. We
may hypothesize, however, that the spread of se/sibi follows two
parallel, distinct paths along the continuum of Unaccusativity/
Unergativity illustrated in table 2.
In particular, sibi seems to proceed from intransitive change of
state/inherently directed change of location verbs to non-directed
change of location verbs and then states, i.e., there is a progression
from high to low Unaccusativity.
Se on the other hand spreads from anticausatives denoting
change of state to states (mainly mental process, though it can
also occur with location state verbs) and then to emotive speech acts
and other activity verbs, i.e., it proceeds from low Unaccusativity to
Unergativity.
By the 8th±9th centuries the two paths converge, and the
distinction between the two sets of intransitives gets formally
neutralized, due to the falling together of the dative and accusative
forms of the reflexive, so that the distinction is lost and se gets used
with unaccusative verbs and sibi occurs with unergative ones (cf.
Norberg 1943: 171±2; DahleÂn 1964, who discuss the issue within a
traditional framework). By this time there is no longer a distinction
between iumentum se refrigerat and iumentum sibi refrigerat, cicatrix
se sanat and cicatrix sibi sanatur, and one can also find se with verbs
that in earlier centuries only occurred with sibi (i.e., also with
unaccusatives) (e.g. se ire `go', se turnare, se reverti `go back'), and
sibi with verbs that most typically took se (e.g., sibi recognoscere
`recall').
This pattern of active alignment continues however in early
Romance, and is still apparent in the Romance languages, though
to a different extent.9
Confining our discussion to Italian, we can notice that, unlike
9
It is also worth noticing the following Spanish examples, where the occurrence vs
non-occurrence of the reflexive with some intransitive verbs appears to be a reflex of
the restrictions we noticed in Late Latin for the occurrence of sibi with intransitive
verbs denoting change of state:
iii. Juan se murioÁ (accidental death) `John died'
iii. Juan se murioÁ *asesinado
iii. Juan murioÁ; asesinado `John was murdered'
4. Conclusion
The area of Late Latin syntax investigated provides further evidence
for the diachronic relevance of the Unaccusative Hypothesis,
pointing to the validity of semantic and syntactico-semantic
approaches to the phenomenon along the lines proposed by Van
Valin 1990; Sorace 1995.
The data so far discussed, in fact, show that at some point in Late
Latin, roughly between the 3rd±4th century A.D., truly pleonastic
Texts
Acc. Trag.= Accius, Oeuvres (fragments), par J. Dangel, Paris: Les Belles Lettres,
1995.
Agnell. = Agnelli Liber Pontificalis Ecclesiae Ravennatis, ed. O. Holder-Egger,
Monumenta Germaniae Historica , Scriptores Rerum Langobardicarum et Italicarum
Saec. VI-IX, I, Hannover: Hahnsche, 1878.
Sen. Thy. = L. Annaeus Seneca, Thyestes, in L. Annaei Seneca, Tragoediae, rec. Otto
Zwierlein, Oxford: Oxford University Press, 19883.
Soran. = Sorani Gynaeciorum, ed. V. Rose, Leipzig: Teubner, 1882.
Sort. Sang.= Die OrakelspruÈche im St. Galler Palimpsestcodex 908 (die sogenannten
Sortes Sangallenses), hrsg. von A. Dold, Wien: O È sterreische Akademie der
Wissenschaft, 1948.
Ter. Ad. = P. Terentius Afer, Adelphoe, in Publi Terenti Afri, Comoediae, rec.
R. Kauer et W. M. Lindsay, Oxford: Oxford University Press, 1926.
Ter. Andr.= P. Terentius Afer, Andria, in P. Terenti Afri, Comoediae, rec. R. Kauer et
W. M. Lindsay, Oxford: Oxford University Press, 1926.
Tert. Nat. = Q.S. Fl. Tertulliani, Ad Nationes, Libri II, c.et s. J. G. Ph. Borleffs,
Turnholt: Brepols, 1954 (Corpus Christianorum 1)
Verg. ecl. = P. Vergilius Maro, Eclogae, in P. Vergili Maronis, Opera, rec. R. A. B.
Mynors, Oxford: Oxford University Press, 1972.
Vulg. act. = Liber Actuum Apostulorum, in Biblia Sacra Iuxta Vulgatam Versionem,
rec. R. Weber OSB, tom. 2, Stuttgart: WuÈrttembergische Bibelanstalt, 19752.
Vulg. Mich.= Micha, in Biblia Sacra Iuxta Vulgatam Versionem, rec. R. Weber OSB,
tom. 1, Stuttgart: WuÈrttembergische Bibelanstalt, 19752.
References
Andrews, Avery, 1985. `The major functions of the noun phrase', in Timothy
Shopen (ed.), Language Typology and Syntactic Description, vol. 1, Cambridge:
Cambridge University Press, 62±154.
Bard, Ellen G., Robertson, Daniel and Sorace, Antonella, 1996. `Magnitude
estimation of linguistic acceptability', Language 72, 32±68.
Bassols de Climent, Mariano, 1945. Syntaxis Historica de la Lengua Latina,vol. 1,
Barcelona: Escuela de FilologõÂa.
Bennett, Charles E., 1914. Syntax of Early Latin, vol.2, Boston: Allyn & Bacon.
Burzio, Luigi, 1986. Italian Syntax, Dordrecht: Reidel.
Cennamo, Michela, 1991. `Se, sibi, suus nelle Inscriptiones Latinae Christianae
Veteres ed i successivi sviluppi romanzi', Medioevo Romanzo 25, 3±21.
Cennamo, Michela, 1993. The Reanalysis of Reflexives: a Diachronic Perspective,
Napoli: Liguori.
Cennamo, Michela, 1995. `Patterns of active syntax in Late Latin pleonastic
reflexives', in John C. Smith and Delia Bentley (eds), Proceedings of the XII
International Conference on Historical Linguistics, Selected Papers, Amsterdam:
Benjamins, to appear.
Cennamo, Michela, 1997. `Tra scritto e parlato: l'extended accusative' e le nozioni di
voce e di relazione grammaticale nel latino tardo e medievale', in Rosanna
Sornicola (ed.), Langue eÂcrite et Langue ParleÂe,, Napoli: Liguori, to appear.
Cennamo, Michela, 1998a. `TransitivitaÁ e inaccusativitaÁ in testi antichi abruzzesi e
napoletani', in Paolo Ramat and Elisa Roma (eds), 93±109.
Cennamo, Michela, 1998b.'The loss of the voice dimension between Late Latin and
early Romance', in Monica Schmid, Jennifer R. Austin and Dieter Stein (eds),
Historical Linguistics 1977: Selected Papers from the XIII International Conference
on Historical Linguistics, DuÈsseldorf 10±17 August 1997, Amsterdam: Benjamins,
77±100.
Cennamo, Michela, 1998c. `Classi verbali e cambiamento sintattico: la reinterpre-
tazione passiva del costrutto riflessivo', paper delivered at the XXXII Convegno
Internazionale della SocietaÁ di Linguistica Italiana, Budapest 29±31 October 1998.
Cennamo, Michela, 1999. `InaccusativitaÁ tardo-latina e suoi riflessi in testi italiani
antichi centro-meridionali', Zeitschrift fuÈr romanische Philologie 115.2.
Cennamo, Michela (forthcoming a). `Transitivity and argument marking in Late
Latin'.
Cennamo, Michela (forthcoming b). `Tracce di inaccusativitaÁ tardo-latina in alcuni
dialetti italiani centro-meridionali'.
Cennamo, Michela (in preparation). Transitivity and Argument Structure: a Dia-
chronic/Variational Study.
Centineo, Giulia, 1986. `A lexical theory of auxiliary selection in Italian, Davies
Working Papers in Linguistics 1: 1±35.
Chierchia, Gennaro, 1992. `Subjects and Aspect', MS.
Compernass, Johannes, 1908. `Vulgaria' Glotta 8: 88±121.
Comrie, Bernard, 1989. Language Universals and Linguistic Typology, 2nd ed.,
London: Blackwell.
DahleÂn , Eric, 1964. Etudes Syntaxiques sur les Pronoms RefleÂchis PleÂonastiques en
Latin, GoÈteborg: Acta Universitatis Gothoburgensis.
DahleÂn, Eric, 1977. Remarques Syntaxiques sur Certains Verbes Pronominaux en
Latin et en Langues Romanes, GoÈteborg: Acta Universitatis Gothoburgensis.
Dixon, David R., 1979. `Ergativity', Language 55, 59±138.
Dixon, David R., 1994. Ergativity, Cambridge: Cambridge University Press.
Dowty, David, 1979. Word Meaning and Montague Grammar, Reidel: Dordrecht.
Durie, Mark, 1987. `Grammatical Relations in Acehnese', Studies in Language 11,
365±99.
Feltenius, Leif, 1977. Intransitivizations in Latin, Uppsala: Almkvist & Wiksell.
Foley, William A, and Van Valin Jr., Robert D., 1984. Functional Syntax and
Universal Grammar, Cambridge: Cambridge University Press.
Grimshaw, Jane, 1987. `Unaccusatives: an overview', Proceedings of the North-East
Linguistic Society 17: 244±58.
Grimshaw, Jane, 1990. Argument Structure, Cambridge MA: MIT Press.
Guerssel, Mohamed, et al., 1985. `A cross-linguistic study of transitivity alterna-
tions', in William H. Eilfort, Paul D. Kroeber and Karen L. Peterson (eds)
Papers from the Parasession on Causatives and Agentivity, Chicago: Chicago
Linguistic Society, 48±63.
Haiman, John, 1983. `Iconic and economic motivation', Language 59, 781±819.
Haiman, John, 1985. Natural Syntax, Cambridge: Cambridge University Press.
Hale, Ken and Samuel J. Keyser, 1986. `Some transitivity alternations in English',
Lexicon Project Working Papers 7, Center for Cognitive Science, Cambridge,
Mass.: MIT.
Hale, Ken and Samuel J. Keyser, 1987 `A view from the middle', Lexicon Project
Working Papers 10, Center for Cognitive Science, Cambridge, Mass,: MIT.
Harris, Alice, 1982. `Georgian and the unaccusative hypothesis', Language 58, 290±
306.
Harris, Alice, 1990. `Alignment typology and diachronic change', in Winfred P.
Lehmann (ed.), Language Typology 1987. Systematic Balance in Language,
Amsterdam: Benjamins, 67±90.
Harris, Martin, 1978. `Alternatives to the morphological passive in Romance',
Semasia 5, 65±87.