Word order variation in Latin verb clusters:
a diachronic perspective
Lieven Danckaert
(UGent, GIST)
1. Introduction
• Empirical focus of the talk is the syntax of objects (O’s), non-finite verbs
(V’s) and auxiliaries (Aux’s) in Latin.
• More specifically, I will look at the diachronic evolution of two so-called
‘head-final sequences’, as in (1):
(1)
legati urbem ingressi sunt.
'The ambassadors entered the city.'
(= Liv. aUc. 45.2.3)
SOVAux
• In this example, two complements precede their heads:
• Head-final sequence I: VP-Aux [TP [VP urbem ingressi] sunt]
• Head-final sequence II: OV [VP urbem ingressi]
1. Introduction
• In classical Latin, both the head-final and the head-initial orders were
available for both structures involved:
(2)
(3)
a. Illae Minoem occiderunt.
'These women killed Minos.' (= Hyg. Fab. 44)
b. etiamsi merito occidit hominem
'even if he was right in killing the man.'
(= Sen. rhet. Contr. 1.2.14)
OV
a. Non est moratus Stichus.
'Stichus did not stay around.' (= Petr. Sat. 78)
b. In uita moratus est.
'He stayed alive.' (= Sen. Ep. 93.3)
AuxVP
VO
VPAux
1. Introduction
• Observation: the structure [[OV]Aux]
• is the statistically predominant pattern in classical Latin, but
• it is lost absent from the modern Romance languages, where
[Aux[VO]] is almost completely generalized
• Aim of the talk: reconstruct part of this evolution, concentrating on the
period 100 BC – 550 AD
• Main empirical result: whereas a clear decline of the order VPAux can
be observed, no such evolution takes place for the pattern OV
• The latter result is at odds with standard claims in the literature (cf.
section 5 below)
Overview
• 1. Introduction: aims of the talk and background assumptions
• 2. Latin clause structure: object positions
• 3. A large-scale corpus study: methodology and corpus description
• 4. Head-final sequence I: VP-Aux
• 5. Head-final sequence II: OV
• 6. Conclusion
2. Latin clause structure
• Background assumption: Latin as a (i) discourse (ii) configurational
language
• Discourse: much of the very flexible properties of Latin word order are
to be explained in terms of information structure
• Rule of thumb: different position => different interpretation
• Configurationality
• Hierarchical structure (brackets, trees), not just the linear string
• For instance: existence of a VP constituent (coordination,
preposing, pronominalization, relativization: cf. handout for some
examples)
• As a result, discourse notions like ‘Topic’ and ‘Focus’ are likely to
be configurationally realized in specialized discourse projections
2. Latin clause structure
• Observation: there is more than one object position
(4)
Non enim a uapore umor corrumpere poterit
SVAuxO
materiem contignationis.
'For the moisture from the heat cannot affect the timbering.'
(= Vitr. Arch. 5.10.3)
(5)
cum testamento scriptus heres euincere
SVOAux
hereditatem possit.
'... since because of the testament the appointed heir can to
recover the heritage.' (= Gai. Inst. 3.36)
(6)
quin seruus beneficium dare possit [...].
‘that a slave.’ (= Sen. Ben. 3.19.1)
(7)
ut nullam calamitatem res publica accipere possit [...]. OSVAux
'so that the state could not suffer any disaster.'
(= Cic. Phil. 7.20)
SOVAux
2. Latin clause structure
• The different object positions we identified could be represented with
the following templates:
(8)
a. OBJ1 S OBJ2
b. OBJ1 S OBJ2
V-fin
V+fin
OBJ3 Aux
OBJ3
• Consider now (9-10):
(9)
Pisum coques.
'Cook the peas.' (= Apic. Res coq. 5.3.3)
(10)
Adicies oleum.
'Add oil.' (= Apic. Res coq. 5.2.2)
OBJ4
OBJ4
2. Latin clause structure
• In the light of the templates in (8), we have to conclude that the
constituency of sentences like (9-10) cannot be determined:
(8)
a. OBJ1 S OBJ2
b. OBJ1 S OBJ2
Aux
V+fin
OBJ3 V-fin
OBJ3
OBJ4
OBJ4
 As a result, sentences like (9-10) should not be taken up in a study
that investigates object positions
 only sentences with a non-finite verb and at least one auxiliary are
really useful: in these, V and T can be told apart, and object positions
can be evaluated with respect to (at least) two reference points
3. A corpus study
• Data from LASLA database (‘Laboratoire d’Analyse Statistique des Langues
Anciennes’, Université de Liège; http://www.cipl.ulg.ac.be/Lasla/index.html)
• Other: www.brepolis.net (BTL)
• In this study: selection of Latin prose texts
• Texts from the classical period not (yet) included in the LASLA corpus
(Vitruvius, Frontinus, Gaius)
• Late Latin texts (from Palladius (ca. 350 AD) to Iordanes (ca. 550 AD))
3. A corpus study
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
Author (work(s))
Date
Cicero (selection of speeches; De officiis,...)
ca. 60 BC
Caesar (De bello ciuili, De bello Gallico I-VII)
ca. 50 BC
Varro (Res rustica; De lingua Latina)
45 BC
Hyginus (Astronomia)
ca. 20 BC
Vitruvius (De architectura)
0 AD
Seneca (Epistulae, Dialogi, ...)
ca. 50 AD
Petronius (Satyricon reliquiae)
ca. 60 AD
Frontinus (Strategemata, De aquaeductu urbis Romae)
ca. 90 AD
Tacitus (Germania, Dialogus de oratoribus, minora)
ca. 110 AD
Gaius (Institutiones)
ca. 170 AD
ca. 170 - 350 AD: very few testimonia
Palladius (De veterinaria medicina, De agricultura)
ca. 350
Itinerarium Egeriae
381-384 AD
Gesta Conlationis Carthaginiensis
411 AD
Vegetius (Epitoma rei militaris, Mulomedicina)
ca. 420 AD
Victor Vitensis (Hist. persec. Africanae prouinciae)
ca. 490 AD
Pompeius Maurus (Commentum Artis Donati)
ca. 500 AD
Caesarius Arelatensis (Sermones 1-80)
ca. 520 AD
Iordanes (Getica, Romana)
ca. 550 AD
Perio
EARLY
EARLY
EARLY
EARLY
EARLY
EARLY
EARLY
EARLY
EARLY
EARLY
Source
Hyperbase
Hyperbase
Brepolis.net
Brepolis.net
Brepolis.net
Hyperbase
Hyperbase
Brepolis.net
Brepolis.net
Brepolis.net
LATE
LATE
LATE
LATE
LATE
LATE
LATE
LATE
Brepolis.net
Brepolis.net
Brepolis.net
Brepolis.net
Brepolis.net
Brepolis.net
Brepolis.net
Brepolis.net
3. A corpus study
• What went in the sample?
• all two- or three-member clusters consisting of (i) a modal auxiliary (in
most cases), (ii) an infinitive, and (iii), when the selected infinitive is
transitive, the complement of this infinitive
• Why auxiliaries that take an infinitival complement?
• frequently used => guaranteed to have a lot of tokens
• relatively easy to find in a corpus
• Which auxiliaries (see handout for details)?
• possum and debeo for long texts
• for shorter texts, possum and debeo, supplemented with uolo, soleo,
incipio, nolo, desino, audeo, conor, malo, in order to at least gather ca. 25
three-member clusters
3. A corpus study
4. Head-final sequence I: VPAux
4. Head-final sequence I: VPAux
Average rates of AuxVP:
- earlier period : 37,85%
- later period: 65,92%
Statistically significant?
Yes (Independent
samples t-test, p = .004)
5. Head-final sequence II: VO
• The data in Ledgeway (2012: ch. 5) suggest the following frequencies for
the order OV (details see handout):
• early Latin (ca. 100 BC – 100 AD, 22 text samples): 73,4%
• late Latin (ca. 350 – 450 AD, 5 text samples): 36,3%
• The difference between those two average frequencies is statistically
significant (t-test for independent samples, p<.001)
• A different picture emerges if we look at object positions in clauses with
an infinitival verb and an auxiliary…
5. Head-final sequence II: VO
5. Head-final sequence II: VO
Average rates of VO:
- earlier period : 26,72%,
- later period: 32,05%
Statistically significant?
No (Independent samples
t-test, p = .449)
6. Discussion
• How can we account for the high frequencies of VO in later texts if clauses
with synthetic verbs are taken into account (cfr. Ledgeway’s data)?
• Not only more than one object position, also more than one position for
finite verbs:
(11)
quin seruus beneficium dare possit [...].
‘that a slave can.' (= Sen. Ben. 3.19.1-2)
SOVAux
(12)
potest mater eius causam probare.
AuxSOV
'his mother is allowed to prove the case.' (= Gai. Inst. 1.32)
• presumably also for lexical (synthetic) finite verbs
(13)
Interfecit Opimius Gracchum.
'Opimius killed Gracchus.' (= Cic. de Or. 2.132)
VSO
6. Discussion
• This gives rise to the following (partial!) linear templates; this is what
we had earlier:
(8)
a. OBJ1 S OBJ2
b. OBJ1 S OBJ2
V-fin
V+fin
OBJ3 Aux
OBJ3
OBJ4
OBJ4
• with additional verb positions:
(14)
a. Aux1 OBJ1
b. V+fin1 OBJ1
S OBJ2
S OBJ2
V-fin OBJ3
V+fin2 OBJ3
Aux2
OBJ4
OBJ4
• Hypothesis: the high verb position is the one that was targeted by finite
verbs in early Romance ‘V2’-clauses (Benincà 1983-'84, Vanelli, Renzi &
Benincà 1985, Adams 1987, Vance 1997, Benincà & Poletto 2010)
6. Discussion
• In addition, the present data confirm the hypothesis that diachronically,
hierarchically higher head-final become head-initial before the head-final
sequences they dominate (cf. Sigurðsson 1988; Biberauer et al. 2010):
7. Conclusion
• The present approach differs in a number of respects from earlier
studies:
1. Size of the investigated corpus
2. Linear order is not taken to be a primitive; constituency
3. Attention is paid to the placement of functional material (in
casu auxiliaries; also possible: negation, adverbs, conjunctions
(cf. Danckaert 2012)
• These factors allow us to obtain a good understanding of diachronic
syntactic facts
• But of course, much more work needs to be done…
7. Conclusion
For future research:
• look at analytic deponent verbs
• enlarge the dataset (Livius, Plinius maior, Plinius minor, Celsus and
Columella for the earlier period; Digesta, Grégoire de Tours and
Fredegarius for the later)
• take into syntactic factors like
• clause type and/or illocutionary force
• embedded vs. main clauses
• type of O (clause, DP, pronoun)
• type of V (participles vs. infinitives)
• type of Aux (BE-auxiliary, different types of modals)
• language-external factors (register, genre,…)
• thorough statistic analysis of the data
Descargar

Dia 1 - UGent