LATIN WORD ORDER

This page intentionally left blank

LATIN WORD ORDER

Structured Meaning and Information

A.M. Devine
Laurence D. Stephens
This page intentionally left blank
 Oxford University Press, Inc., publishes works that further
Oxford University's objective of excellence
in research, scholarship, and education.
Oxford New York
Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto
With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam
Copyright © 2006 by A.M. Devine and Laurence D. Stephens
Published by Oxford University Press, Inc.
198 Madison Avenue, New York, New York 10016
www.oup.com
Oxford is a registered trademark of Oxford University Press
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means.
electronic, mechanical, photocopying, recording, or otherwise,
without the prior permission of Oxford University Press.
Library of Congress Cataloging-in-Publication Data
Devine, A.M. (Andrew M.)
Latin Word Order : structured meaning and information /
A.M. Devine and Laurence D. Stephens.
p. cm.
Includes bibliographical references and index.
ISBN-13 978-0-19-518168-5
ISBN 0-19-518168-9
1. Latin language—Word order. 2. Latin language—Semantics.
1. Stephens, Laurence D. II. Title.
PA2293.D48 2006
475—dc22 2005054651
135798642
Printed in the United States of America
on acid-free paper
 PREFACE
Word order is not a subject anyone reading Latin can afford to ignore: apart from anything else, word order is what gets one from
disjoint sentences to coherent text. Reading a paragraph of Latin without attention to the word order entails losing access to a
whole dimension of meaning, or at best using inferential procedures to guess at what is actually overtly encoded in the syntax.
This book begins by introducing the reader to the linguistic concepts, formalism and analytical techniques necessary for the study
of Latin word order. It then proceeds to present and analyze a representative selection of data in sufficient detail for the reader to
develop both an intuitive grasp of the often rather subtle principles controlling Latin word order and a theoretically grounded
understanding of the system that underlies it. Combining the rich empirical documentation of traditional philological approaches
with the deeper theoretical insight of modern linguistics, our work aims to reduce the intricate surface patterns of Latin word
order to a simple and general crosscategorial system of syntactic structure which translates more or less directly into constituents
of pragmatic and semantic meaning.
Two OUP reviewers read the manuscript in early 2004, and we would like to take this opportunity to thank them both for their
helpful suggestions at that stage in the project. We would also like to express our gratitude to Susan F. Stephens for her help with
the translation of the examples. Errors remain our responsibility.
A.M.D., L.D.S.
 CONTENTS
ABBREVIATIONS xi
Introduction 3
Semantics and pragmatics 3
Studying word order 5
Thematic Roles 9
Pragmatic values 13
Notation for semantic and pragmatic meaning 17
Syntactic structure 25
Prosodic structure 29
Postscript 30
1 Arguments of Verbs 36
1.1 SUBJECTS 37
1.2 DITRANSITIVES 40
Dare 40
Other verbs of giving 45
Donare 48
Adimere 49
Extorquere and eripere 51
Prepositional prefix verbs 52
1.3 OBLIQUE ARGUMENTS: LOCATIVES AND INSTRUMENTS 55
Conlocare 55
Liberare 56
Instrumental arguments 57
Goal phrases 58
Source phrases 63
1.4 ADJUNCTS 64
Instrumentals 64
Causa 67
Time and place adverbials 69
Manner and means phrases 71
Comitatives 75
Ablative absolutes 77
1.5 STRUCTURAL ANALYSIS 79
Neutral order 79
Flat syntax 82
Complement Syntax 84
Specifier syntax 87
Postscript 97
1.6 SCRAMBLING 98
Adverbs of manner 101
Structural analysis 108
Obliques in Cato 112
1.7 POSTVERBAL CONSTITUENTS 119
Postverbal constituents in Caesar 119
V-bar syntax 125
2 Verb Positions 145
2.1 VERB INITIAL 145
Polarity focus 145
Imperatives 149
Existential and presentational sentences 150
Passives 154
Psych verbs 154
Discourse cohesion operators 157
Conjoined structures 163
Structural analysis 166
2.2 VERB SECOND 172
Negative quantifiers 172
Weak focus 176
Structural analysis 179
2.3 THE AUXILIARY 179
viii

Eventive sentences 180

Raising to focus 181
Negation 183
Subordinate clauses 184
Individual verbs 187
Structural analysis 191
2.4 THE COPULA 198
Relative subjects 198
Lexical subjects 199
Negation 202
Raising to CP 202
Structural Analysis 204
2.5 EXISTENTIAL ESSE 208
Structural analysis 213
2.6 SALLUST AND CATO 216
Sallust 216
Cato 218
3 Strong and Weak Arguments 225
3.1 TOPIC AND FOCUS 225
Strong narrow focus 225
Yes-No questions 235
Nonne questions 237
Alternative questions 240
Chiasmus with arguments and adjuncts 242
Chiasmus with a verb 246
3.2 ASSOCIATION WITH FOCUS 249
Semper, saepe 250
Numquam 258
Non solum, non modo 262
Quoque 265
Ne quidem 266
Split proper names 272
Structural analysis 275
3.3 WEAK PRONOUNS 277
Strong and weak pronouns 277
Weak pronoun positions 282
Weak pronouns after Comp 287
Domain external structures 292
Raising to focus 294
Structural Analysis 298
Weak pronouns and modified noun phrases 303
4 Arguments of Nominals 314
4.1 NOUN PHRASES 315
Subjective and objective genitives 316
Psych nouns 317
The left node rule: Erga complements 331
Nomina actionis 337
Nomina agentis 343
Relational nouns 352
Possessives 361
Partitives 368
Narrow strong focus in noun phrases 376
4.2 STRUCTURAL ANALYSIS 377
Structures for genitives 378
Theories for Latin 380
Posthead genitives 381
Prehead genitives 383
The functional theory 384
Cross-categoriality 385
Focus semantics in NPs 386
Informational individuation 387
The prosodic theory 389
4.3 ADJECTIVE PHRASES 391
Immemor 391
Expers 393
Cupidus 396
Structural analysis 398
 ix

5 Modiers 403
5.1 INTERSECTIVES 405
Adjectives of material 405
Adjectives in -arius 413
Adjectives of season 417
Inherent properties 419
5.2 ADJECTIVES FROM PROPER NAMES 426
Geographical locations 426
Names of subkinds 430
Personal name adjectives in Caesar 430
Personal name adjectives in Cicero 432
Bellum with proper name adjective 439
Appendix: Bellum civile 442
5.3 AGE AND EVALUATION 446
Vetus 446
Novus 449
Superior 452
Gravis 456
Mirificus, pestifer 457
Nefarius 460
Perditus homo 464
Modified homo 465
5.4 MEASURE; STACKED ADJECTIVES 467
Parv(ul)us 467
Magnus 471
Maximus 474
Stacked adjectives 476
5.5 STRUCTURAL ANALYSIS 481
Theories for Latin 483
Postscript 491
5.6 QUANTIFIERS, DEMONSTRATIVES 492
Cardinals 492
Many, much 498
Universal quantifier 507
Demonstrative adjectives 511
6 Hyperbaton 524
6.1 GENITIVE HYPERBATON 525
6.2 POSTMODIFIER HYPERBATON 531
6.3 PREMODIFIER HYPERBATON 540
Premodifier hyperbaton 540
Pragmatics of premodifier hyperbaton 542
Syntax of premodifier hyperbaton 548
Premodifier hyperbaton in subordinate clauses 559
6.4 HYPERBATON WITH NONVERBAL HEADS 563
Hyperbaton with nominal heads 563
Hyperbaton with prepositional heads 568
Attribute complement hyperbaton 575
6.5 MINOR HYPERBATA 578
Degree Phrase hyperbaton 578
Clitic hyperbaton 583
Interrogative hyperbaton 584
Conjunct hyperbaton 586
6.6 STRUCTURAL ANALYSIS 591
Variation across styles 602
V-bar syntax and hyperbaton 604
Bibliography 611
Index nominum 631
Index rerum 636
This page intentionally left blank
 ABBREVIATIONS
Acad Cicero Academica
Ad Att Cicero Epistulae ad Atticum
Ad Brut Cicero Epistulae ad Brutum
Ad Fam Cicero Epistulae ad Familiares
Ad Qfr Cicero Epistulae ad Quintum fratrem
Aen Vergil Aeneid
Aul Plautus Aulularia
Aul Gell Aulus Gellius
Bacch Plautus Bacchides
BAfrDe Bello Africo
BAlexDe Bello Alexandrino
BC Caesar De Bello Civili
BG Caesar De Bello Gallico
BHispDe Bello Hispaniensi
Brut Cicero Brutus
Capt Plautus Captivi
Cat Cicero In Catilinam
Cato Cato De Agri Cultura
Cato Orat Cato Speeches
Cato Orig Cato Origines
CIL Corpus Inscriptionum Latinarum
Col Columella De Re Rustica
Col De Arb Columella De Arboribus
De Amic Cicero De Amicitia
De Div Cicero De Divinatione
De Dom Cicero De Domo Sua
De Fat Cicero De Fato
De Fin Cicero De Finibus
De Har Resp Cicero De Haruspicum Responso
De Inv Cicero De Inventione
De Leg Agr Cicero De Lege Agraria
De Leg Cicero De Legibus
De Nat Deor Cicero De Natura Deorum
De Off Cicero De Officiis
De Orat Cicero De Oratore
De Part Or Cicero De Partitione Oratoria
De Prov Cicero De Provinciis Consularibus
De Rep Cicero De Republica
De Sen Cicero De Senectute
Orat Cicero Orator
Div Caec Cicero Divinatio in Q. Caecilium
Georg Vergil Georgics
Hor Sat Horace Satires
In Pis Cicero In Pisonem
In Vat Cicero In Vatinium
Jug Sallust Jugurtha
Luc Cicero Lucullus
Lucr Lucretius
Ov Ex Pont Ovid Ex Ponto
Ov Fast Ovid Fasti
Ov Her Ovid Heroides
Ov Met Ovid Metamorphoses
Ov Trist Ovid Tristia
Parad Cicero Paradoxa Stoicorum
Phil Cicero Philippics
Phorm Terence Phormio
Plaut Stich Plautus Stichus
Plaut Amph Plautus Amphitruo
Plaut Curc Plautus Curculio
Plaut Truc Plautus Truculentus
xii

Plaut Mil Plautus Miles Gloriosus

Pliny NH Pliny Naturalis Historia
Pliny Paneg Pliny Panegyricus
Post Red Pop Cicero Post Reditum ad Populum
Post Red Sen Cicero Post Reditum in Senatu
Pro Arch Cicero Pro Archia
Pro Balb Cicero Pro Balbo
Pro Caec Cicero Pro Caecina
Pro Cael Cicero Pro Caelio
Pro Clu Cicero Pro Cluentio
Pro Flacc Cicero Pro Flacco
Pro Font Cicero Pro Fonteio
Pro Lig Cicero Pro Ligario
Pro Leg Man Cicero Pro Lege Manilia
Pro Marc Cicero Pro Marcello
Pro Mil Cicero Pro Milone
Pro Mur Cicero Pro Murena
Pro Planc Cicero Pro Plancio
Pro Quinct Cicero Pro Quinctio
Pro Rab Perd Cicero Pro Rabirio Perduellionis Reo
Pro Rab Post Cicero Pro Rabirio Postumo
Pro Reg Cicero Pro Rege Deiotaro Deiot
Pro Rosc Am Cicero Pro Roscio Amerino
Pro Rosc Com Cicero Pro Roscio Comoedo
Pro Scaur Cicero Pro Scauro
Pro Sest Cicero Pro Sestio
Pro Sull Cicero Pro Sulla
Pro Tull Cicero Pro Tullio
Rhet HerRhetorica ad Herennium
Sall Cat Sallust Catilina
Sall De Rep Sallust De Republica
Sall Hist Sallust Histories
Sen Contr Seneca Controversiae
Sen De Ben Seneca De Beneficiis
Sen Ep Seneca Epistulae
Tac Ann Tacitus Annals
Tac Ger Tacitus Germania
Tac Hist Tacitus Histories
Top Cicero Topica
Tusc Cicero Tusculanae Disputationes
Varr LL Varro De Lingua Latina
Varro RR Varro Res Rusticae
Verr Cicero In Verrem
Vitruv Vitruvius
 INTRODUCTION

Omnis Gallia in tres partes divisa est

Semantics and pragmatics

Consider the following dialogue between an instructor (A) and two of his students (B, C)
A. What happened in the senate1 on the Ides of March 44 B.C.?
B. Napoleon stabbed Mrs Thatcher.
C. Brutus did stab Caesar. In the senate it happened. It was Cassius that stabbed him.
B's answer is pragmatically impeccable, by which we mean that he presents the information in a way that fits
congruently and coherently into the discourse. He was asked to describe an event, and that is what he does. The
problem with B's answer involves semantics: he has asserted as true a proposition that is false. C on the other hand
gets his facts right: Brutus and Cassius stabbed Caesar in the senate. The problem with C's answer lies in the area of
pragmatics:2 his answer is incongruous because he does not structure the information appropriately. His first sentence
presents new information as though it were established information that has been called into question. His second
sentence presents established information as new. His third sentence excludes information that has been previously
asserted. The result is an entirely incoherent text.
Now suppose the instructor3 asks B and C to translate four Latin sentences on the same familiar topic of Roman
history
Caesarem Brutus occidit
Brutus Caesarem occidit
Occidit Brutus Caesarem
Caesarem occidit Brutus.
The semantic meaning encoded by the morphology is the same in all four sentences, and accordingly the morphology
does not vary from one sentence to another. Since the lexical meanings do not vary either, the sentences are logically
4 INTRODUCTION

(truthconditionally) equivalent.4 However, the pragmatic (informational) meaning, which is encoded by the word order,
does vary from one sentence to another, and accordingly the sentences have different word orders. B, as you may
suspect, is not a very good student. Not only does he have rather hazy notions of European history, his knowledge of
Latin nominal morphology is limited to the first declension. He can only guess at the grammatical relations in these
sentences, and, regrettably, decides on ‘Caesar killed Brutus.’ C on the other hand has no problem with Latin
morphology; he knows all five declensions and accesses the correct semantics immediately. But as we have seen he is
pragmatically challenged. Latin texts for him are a sort of word salad that has to be untossed before it can be
translated. He assumes that the four sentences have identical pragmatic meanings. Or perhaps he suspects that they are
different, but can only guess what information each sentence would contribute to a discourse context and in what sort
of context each sentence would be felicitous. In concrete terms, for C the following are simply equivalent
What happened was that Brutus killed Caesar
It was Caesar that Brutus killed
It was Brutus that killed Caesar
Who Brutus killed was Caesar
What happened to Caesar was that Brutus killed him
Brutus actually did kill Caesar
As far as Caesar is concerned, Brutus killed him
What Brutus did to Caesar was kill him.
Speakers do not present information to their interlocutors in a haphazard or unstructured manner, higgledy-piggledy.
They cooperate by packaging it in such a way that the hearer is able to integrate it into the progress of the ongoing
discourse. That way the hearer does not need to expend time and effort trying to figure out what the speaker is talking
about and what new information he is trying to contribute. A set of instructions is encoded into the message, which
hearers use to decode its pragmatic meaning, just as they use other structural elements of the message to decode its
semantic meaning. The paraphrases we cited above employed rather marked constructions to convey distinctions of
pragmatic meaning in written English. The normal unmarked way to encode pragmatic meanings in spoken English is
via the prosody (stress, intonation, timing, etc.)
Brutus KILLED Caesar
He didn't say Brutus killed CAESAR, he said BRUTUS killed Caesar.
(Read caps as strong stress.) If we transcribe spoken English into regular written text without special typographic
highlighting devices, we lose this component of meaning
He didn't say Brutus killed Caesar, he said Brutus killed Caesar.
 INTRODUCTION 5

Sometimes a pragmatic distinction carries a semantic distinction along with it, so we lose that too
He always took Diana to the theatre
I'm surprised that Jack likes Statius
She only likes Latin verse.
With strong stress on Diana, the first example means that on all occasions in which he took a girlfriend to the theatre,
the girlfriend was Diana; with strong stress on theatre, it means that on all occasions that he took Diana somewhere on
a date, it was to the theatre. In the second example strong stress on Jack gives the reading ‘It surprises me that Jack of
all people should like Statius’; strong stress on Statius gives the reading ‘It surprises me that Jack should like Statius of
all authors.’ If the third example has strong stress on Latin, it excludes her liking Greek verse; if it has strong stress on
verse, it excludes her liking Latin prose (and has nothing to say about her attitude to Greek verse). While such
distinctions cannot be retrieved from the written record of the utterance in English
I said that she only likes Latin verse, not that she only likes Latin verse,
they are, for the most part, readily available in written Latin text, since Latin uses word order (as well as, or perhaps
aligned with, prosody) to encode them.5 Fortunately, the dead language uses word order and the living language, by and
large, does not, rather than the other way around. In any case, speakers of both languages go out of their way to
encode pragmatic meanings overtly into their messages by the consistent application of an elaborately structured
semiotic system, and it is not advisable to make a habit of discounting, filtering out or just ignoring part of the meaning
that an author has built into his text. Reading a paragraph of Latin without attention to the word order is like taking a
black and white photograph. Adding in the word order is like going from black and white to full colour. A whole new
dimension of meaning is added, affording an explicit mapping from syntax into coherent text and revealing a rich
range of subtle interpretive nuances.

Studying word order

Word order is what gets the reader of Latin from disjoint sentences to coherent and incrementally interpretable text. In
this sense, the study of word order is of real practical value to anyone who uses texts as a source of information about
Roman culture, history or literature.6 The reader will soon become aware that the subject is quite challenging. On the
one hand the attested word order patterns are intricate. All the native speakers are dead, and we have to rely on a
corpus of texts transmitted to us by copyists who can have an unhelpful tendency to “normalize” word order.7
Consequently each pattern needs to be substantiated by full exemplification. What never occurs is not necessarily
ungrammatical, it could just be accidentally absent from the corpus; and what
6 INTRODUCTION

occurs just once is not necessarily grammatical, it could merely be an error in transmission. On the other hand, despite
many remarkable observations and intuitions, the techniques of traditional grammar are not sophisticated enough to
produce an explicit and coherent theory of the interface between syntax and pragmatics in Latin. Our aim should be
not just to document the facts, but to understand them, to reduce the kaleidoscopic surface complexity of Latin word
order to a relatively simple and coherent system of general rules. To that end we will need to combine the rich
empirical documentation of nineteenth-century philology with the deeper and more explanatory insights of twentieth-
century theoretical linguistics. If there are no data, there cannot be any theory. If there is no theory, there can hardly be
any understanding. It is true that the resulting analysis is more technical than the comparatively brief remarks on word
order in our standard classroom grammars like Gildersleeve & Lodge or Allen & Greenough. But remember that
those books were first published around the time of the Franco-Prussian war, and the editions we use today are still
over a hundred years old. So it is not surprising if, like most subjects, the study of word order has become more
technical over the ensuing century.
In each chapter, we start by collecting the different word order patterns attested in classical Latin prose and analyzing
their pragmatic and semantic properties. Then we proceed to elicit the syntactic structures associated with the various
patterns, and to consider how these syntactic structures might translate into constituents of semantic and pragmatic
meaning. This procedure entails a decomposition of the traditional one-dimensional question “What does this sentence
mean?” into a multidimensional analysis permitting questions like the following to be addressed as the need arises:8
(i) What is the proposition that the speaker is asserting as true, or what is the situation out there in the world that the
sentence is designed to describe? (semantic meaning)
(ii) What information is the hearer assumed to have before the sentence is uttered, and how does the sentence update
the hearer's information state? (pragmatic meaning)
(iii) What aspects of the syntax of the sentence encode its semantic meaning?
(iv) What aspects of the syntax of the sentence encode its pragmatic meaning?
(v) How is the semantic and pragmatic meaning of the sentence composed out of the meaning of its parts?
(vi) How are these aspects of the syntax processed and interpreted by the hearer as he receives the message in real
time?
Question (v) presupposes that pragmatic meaning, like semantic meaning, is built up compositionally. This is the
central premise of our analysis: pragmatic meaning is compositional, not atomistic. It is empirically obvious that
components of pragmatic meaning can be structurally separated: dislocated topics and tails are iconically9 segregated
by both syntax and prosody, and it is very reasonable to assume similar functional factors to explain the syntax and
prosody
 INTRODUCTION 7

of lower level topical material like subjects and scrambled phrases. The effect of this type of segregation is that the
meaning of an individual pragmatic structure can be composed before it is combined with the meaning of its
neighbour. In a sentence such as Jack likes Tacitus, we recognize two grammatically defined syntactic constituents (the
subject Jack and the verb phrase likes Tacitus); each grammatical constituent translates into a constituent of semantic
meaning; so the meaning of the verb combines with the meaning of the object to give a complex meaning for the verb
phrase, which we can represent in symbols as λx. Like (x, Tacitus). Correspondingly, in a sentence such as Tacitus Jack.
likes,10 we recognize two pragmatically defined syntactic constituents (the topic Tacitus and the comment Jack likes); each
pragmatically defined constituent translates into a constituent of pragmatic meaning; so the meaning of the subject
combines with the meaning of the verb to give a complex meaning for the comment, which we can represent as λy.
Like (Jack, y). That is why the subtitle of this book is ‘Structured Meaning11 and Information.’ If we thought that
pragmatic meaning involved no more than discrete pragmatic values or features, that are read off concatenated
individual words or linear syntactic positions (X < Y < Z), we would have used a subtitle such as ‘Serial Order and
Information.’
It follows that we will be using a relatively multifaceted methodology, combining philological data collection with
theoretical linguistic analysis, and within the latter sphere attempting to interface syntactic analysis with semantic
interpretation, since we are interested not only in which syntactic structure correlates with which pragmatic meaning
but also in the mechanisms by which one is translated into the other. Omission of any of these components would
have risked ending up with a partial rather than a comprehensive understanding. Different parts of the job call for
different tools: the subject matter just does not permit us to choose between philology and linguistics, or between a
formalist syntax and a functionalist pragmatics.12 Each perspective makes its own specific and significant contribution.
Here is a case, if there ever was one, for ignoring the partitions and demarcations of traditional academic disciplines
and subdisciplines. The downside is that, depending on their individual fields of specialization, different readers may be
unfamiliar with, or even unsympathetic to, different aspects of this methodology. Readers with a primarily philological
background will probably be unaccustomed to the general theoretical stance of the discussion, and, conversely, pure
theoreticians may become impatient with the rich philological documentation. Those whose interest is primarily in the
pragmatics may be surprised at the attention devoted to syntactic theory, and vice versa, and both types of reader may
normally pay less attention to semantic issues than we are going to do.
Nevertheless, while the mix of methodologies may be atypical, there is nothing unorthodox about the way in which any
individual methodology is applied. For instance, for our syntax we use a standard syntactic lingua franca (essentially a
sanitized version of pre-Minimalist syntax), with the addition of
8 INTRODUCTION

discourse functional projections.13 For the semantics we utilize two well established formal approaches to
informational structure (tripartite structures and structured meanings). Other theories, both syntactic and semantic, are
available, some of them widely used and some of them far more elaborate and sophisticated, but not all of them are
appropriate for our Latin data. For instance, on the syntactic side, some of the more abstract theories are designed to
provide a rather deep interface between the syntax and the semantics, while the sort of word order variation that is the
subject of this work is more a matter of surface level syntax. The fact that it is sometimes relegated to a post-syntactic,
stylistic (or even prosodic) component of the grammar is evidence of a sensitivity to this distinction. On the semantic
side, theories designed to explain focus in situ are evidently unsuitable for structures in which focus moves, as it often
does in Latin.
While the general target of our work is the simple sentence in Classical prose (the word order of verse is clearly a
separate, though not unrelated, question), we have felt free to pick and choose both texts and lexical material suitable
for our purposes. The primary aim in each case was to find data that might clarify the theoretically significant issues,
not to attempt broad coverage of the subject matter, whether relative to periods, registers and styles or relative to the
range of available lexical material. Consequently our analysis proceeds mainly on the basis of probes or tests, using just
a few representative words14 in a restricted corpus of texts to build the elements of a general theory of Latin word
order. Within those limits, it is often possible to give a quite complete account including finegrained effects. (The latter
cannot be ignored, since minor philological observations can have major theoretical import.) It seemed sensible at this
stage to gravitate towards the clearest evidence and to leave for future research data involving a less transparent
calculus of conditioning factors. Most of the time, the sets of examples cited serve to motivate the theory and illustrate
its application to the data. They are not designed to prove unequivocally the validity of the theory we develop nor the
invalidity of all other possible theories. Since the theory is built up progressively, conclusions at one point in the
discussion depend upon what has been proposed at some earlier point. So for instance if you do not accept that there
is such a thing as neutral word order in Latin, the question of whether Latin is a scrambling language does not even
arise, and the examples cited in the section on scrambling have to be analyzed in some alternative theoretical
framework, for example linear ordering of pragmatic categories. And if you limit your toolbox to just linear ordering,
then the concept of string vacuous scrambling (AB → Ai [—i B]) will strike you as circular and vacuous indeed. (Yet
string vacuous movement is not always empirically unverifiable, since it can have prosodic repercussions.)
The remainder of this introduction is devoted to a survey of the theoretical concepts used thoughout the book along
with the associated formalism and notation.
 INTRODUCTION 9

Thematic Roles
The situations we use simple sentences to describe have been compared to scenes in a “little drama”15 which is part of
our experience and which we are communicating to others. The event or state comprising the scene is typically
encoded by the verb, and the participants in the event or state are encoded by the arguments of the verb, typically noun
phrases or pronouns
1
Baebius exercitum M. Pinario… tradiderat (Livy 40.25.8).
There had been an event of handing over, and the participants in that event (the role players in the little dramatic
scene) had been Baebius, Pinarius and the army. Baebius had played the role of transferrer, the army of transferee, and
Pinarius of recipient of the transferred entity. The roles that the participants play depend upon the scene they are
players in, therefore on the verb. A runner is not playing the same role as a walker, nor is a jogger a sprinter. Someone
who crumples a sheet of paper is not playing the same role as someone who wrinkles, folds or creases it. The semantic
distinction between these roles is lexically encoded by the verb. If it were morphologically encoded too, there would be
a different nominative case for the subject of each verb (which would make the language unwieldy and difficult to
learn). But the semantic distinction between the role of crumpler and that of crumplee does get encoded on the noun
by the morphology (in Latin by the nominative and accusative endings, respectively). In some languages pronominal
elements distinguish between voluntary and involuntary agents: for instance in a Californian Indian language
(Northern Pomo) different pronouns are used for ‘he belched (involuntarily)’ and ‘he belched (on purpose)’. Tsova-
Tush (a Northern Caucasian language) makes a similar distinction for getting drunk, and uses different pronouns for
activities like run versus states like be ripe. How is it that the grammar “sees” certain distinctions between participant
roles and ignores others? Sticking with our theatrical analogy, recall the old joke that all Italian operas have the same
plot: the tenor (tall, handsome and heroic) is in love with the soprano (tall, beautiful and aristocratic); the baritone (not
a nice person) is also in love with the soprano, and/or the mezzo-soprano with the tenor; the bass offers occasional
paternal advice. These generalizations are more or less valid so long as one abstracts away from irrelevant details, such
as whether the tenor is a Roman proconsul, an Egyptian general or a bandit from Aragon. In much the same way, the
grammar abstracts away from the detail of different events and states. It is not concerned with the difference between a
runner and a jogger; it considers them both simply agents in an event. It sees particular scenes as instances of one of a
few classes of more general prototypical scenes or “frames”16 with a limited number of canonical participants. The
latter we call thematic roles.

1
Baebius had handed over the army to M. Pinarius (Livy 40.25.8).
10 INTRODUCTION

Thematic roles are grammatically relevant classes of participant roles. Here are some of the more important ones, with
informal definitions
AGENT: volitional, energized initiator of an action
2
Patrem occidit Sex. Roscius (Pro Rosc Am 39)
EXPERIENCER: one undergoing an involuntary mental state arising from some external stimulus
3
Quis vocem praeconis audivit? (Phil 2.103)
PATIENT: one undergoing and being affected or changed by an action
4
Patrem occidit Sex. Roscius (Pro Rosc Am 39)
STIMULUS: nonvolitional source of a mental state in an experiencer
5
Quis vocem praeconis audivit? (Phil 2.103)
INSTRUMENT: something used by the agent in an action
6
ne cum improbis boni ferro dimicarent (De Dom 5)
THEME: spatially located or displaced entity
7
tela intra vallum coniciebant (BG 5.57)
POSSESSOR: one who possesses some property
8
Fundum habet in agro Thurino M. Tullius paternum (Pro Tull 14)
POSSESSED: property possessed
9
Fundum habet in agro Thurino M. Tullius paternum (Pro Tull 14)
RECIPIENT, DEPRIVEE: one involved in the transfer or property
10
arma Satricanis ademit (Livy 9.16.10)
BENEFACTIVE, MALEFACTIVE: one benefiting or being injured by the action of another
11
Neque enim solum nobis divites esse volumus sed liberis, propinquis, amicis, maximeque rei publicae (De Off 3.63)

2
Sex. Roscius killed his father (Pro Rosc Am 39).
3
Who heard the voice of the auctioneer? (Phil 2.103).
4
Sex. Roscius killed his father (Pro Rosc Am 39).
5
Who heard the voice of the auctioneer? (Phil 2.103).
6
So as to avoid having good men come into armed conflict with wicked men (De Dom 5).
7
Hurled their weapons inside the rampart (BG 5.57).
8
M. Tullius possesses a farm inherited from his father in the territory of Thurium (Pro Tull 14).
9
M. Tullius possesses a farm inherited from his father in the territory of Thurium (Pro Tull 14).
10
He deprived the Satricans of their arms (Livy 9.16.10).
11
For we don't desire to be rich only for ourselves, but also for our children, our relatives, our friends, and above all for the state (De Off 3.63).
 INTRODUCTION 11

COMITATIVE: spatially associated with the agent or participating with the agent in the action
12
Cursare iste homo potens cum filio blando et gratioso circum tribus (Verr 1.25)
LOCATIVE: static location
13
suo stare loco (Livy 9.37.3)
GOAL: location towards which
14
Mancinus domum revenisset (De Orat 1.181)
SOURCE: location from which
15
quem numquam incursiones hostium loco movere potuerunt (Pro Rab Perd 36).
There are a number of other thematic roles, but that's enough to be getting on with.
Different inflectional and/or prepositional coding is often associated with the different roles, as illustrated by many of
the examples. We do not expect to find a difference between one who wounds (vulnerat), one who slaughters (obtruncat)
and one who cuts the throat (iugulat), and in fact they all take agent arguments. But we are prepared for a potential
difference to surface between agents, experiencers and possessors, and in fact we find one: experiencers and possessors
are sometimes coded by an oblique (dative) case, while agents are always in the nominative case in active sentences
16
Placent vobis hominum mores? (Verr 2.3.208)
Quo minus igitur honoris erat poetis (Tusc 1.3)
*Pater occidit Sex. Roscio.
The socalled dative of the agent (rather than ab with the ablative) is found with passive verbs of experience and
cognition
17
Cui non sunt auditae Demosthenis vigiliae? (Tusc 4.44).
Nevertheless most experiencers are coded as subjects; in this condition the grammar declines to make a distinction
between an agent and an experiencer. This neutralization is common across languages. For instance, in the Australian
aboriginal language Warlpiri the agent and the perceiver both appear in the ergative case in the following sentences

12
This powerful man rushed around the tribes with his charming and obliging son (Verr 1.25).
13
They stayed in their positions (Livy 9.37.3).
14
Mancinus had returned home (De Orat 1.181).
15
Who the attacks of the enemy were never able to shift from his position (Pro Rab Perd 36).
16
Are you satisfied with the way men behave? (Verr 2.3.208) So the less esteem poets had (Tusc 1.3).
17
Who hasn't heard of the sleeplessness of Demosthenes? (Tusc 4.44).
12 INTRODUCTION

The man-ERG is spearing the kangaroo

The kangaroo-ERG sees the man.
Just as there was neutralization of distinctions in the mapping from participant roles to thematic roles, so there is
further neutralization in the mapping from thematic roles to grammatical relations and oblique case functions
(otherwise there would be the same number of cases as there are thematic roles, which again would lead to an unwieldy
morphology). The ablative for instance is notoriously polysemous, including the mutually contradictory functions of
locative and source (stare loco, movere loco): you can't both be in a place and be moving away from it at the same time.
Disambiguation comes from prepositions or from the verb or from both.
The linking of thematic roles with grammatical relations is a difficult area of research. Mapping to the subject seems to
be based on a hierarchy such as the following17
Agent > Benefactive > Recipient > Patient/Theme > Location.
The highest (leftmost) argument projected by any verb becomes the subject. However in Latin prose, as in some other
languages, an instrument tends not make a good subject of a transitive clause even when no higher thematic role is
available (An arrow killed King Harold).
The arguments of a verb are those thematic roles that are obligatorily projected by the verb into the syntax. Their
number varies according to the meaning of the verb; the number of arguments projected is called the valency of the
verb.18 Weather verbs like pluit ‘it rains’ project no arguments, but merely describe the event. Unergative intransitive
verbs like rideo ‘I laugh’ project one argument, which is an agent; unaccusative intransitives like morior ‘I die’ project one
argument, which is a patient. Other intransitives additionally project oblique arguments, for instance Romam pervenit ‘he
arrived at Rome,’ patri persuasit ‘he persuaded his father.’ Transitive verbs project two arguments, for instance the agent
and patient in Rabirius Saturninum occidit ‘Rabirius killed Saturninus.’ Ditransitive verbs project three arguments, for
instance the agent, theme and deprivee in Populus Romanus municipiis civitatem ademit ‘The Roman people deprived
municipalities of citizenship.’ Valency can vary: for instance, liberare ‘set free’ can project two arguments (nominative
and accusative) or three (nominative, accusative and ablative). Detransitivization is a common source of variation: eam
esto ‘eat it,’ esse nollent ‘they refuse to eat.’ Verbs can also take additional optional arguments: rident ‘they laugh,’ rident verba
latina ‘they laugh at the Latin language.’ Null (prodropped) arguments (amat ‘he loves her’) have nothing to do with
variation in valency.
Roles that are instantiated in the sentence but are not obligatorily projected are called adjuncts. It is sometimes difficult
to decide whether a phrase is an argument or an adjunct. Time and place expressions can be arguments: habitabat
Athenis ‘he lived at Athens,’ centum viginti annos vixit ‘he lived a hundred
 INTRODUCTION 13

and twenty years,’ but they are usually optionally added to the clause as circumstantial adjuncts: Athenis nuper est mortuus
‘he recently died at Athens.’ Note that while adjuncts are syntactically optional, they can be semantically obligatory: an
event has to occur at some time in some place. Passivization involves the demotion of an agent argument to adjunct
status and the promotion of the patient argument to subject: Rabirius Saturninum occidit – Saturninus (a Rabirio) occisus est.
A patient can also be the subject of an active sentence: Lentulus grave volnus accepit ‘Lentulus suffered a serious wound.’

Pragmatic values
Although people sometimes speak purely for the pleasure of hearing their own voice, the usual reason for uttering a
sentence is to transmit information that you have and that you know or assume your interlocutor does not have. In
discourse, this may be by way of answer to an explicit question, as in comedy
18
A. Quis hic loquitur? B. Sophrona (Phorm 739)
A. Sed quem quaeritas? B. Bacchidem. (Bacch 587)
A. Cuius ducit filiam? B. Vicini huius Euclionis senis (Aul 289: app. crit.)
A. Quid is fecit? B. Confutavit verbis admodum iratum senem (Phorm 477)
A. Quid fecisti? B. In lapicidinas compeditum condidi (Capt 944)
A. Quid ais? B. Huius patrem vidisse me (Phorm 199)
A. Qui non potest? B. Quia uterque utriquest cordi. (Phorm 799).
The qu-word in each question represents the information that A does not have, while the rest of the question
represents information that A presupposes and assumes is presupposed also by his interlocutor B. The answers take
the form of fragmentary sentences. They fill in the requested new information and ellipse the mutually presupposed
old information. In the first two examples (Phorm 739; Bacch 587) the new information is the subject and direct
object phrase respectively. In the third example (Aul 289) it is a possessive, in the fourth (Phorm 477) and fifth (Capt
944) it is the whole verb phrase: In the last two examples the new information is a complete clause: a complement
clause (Phorm 199; accusative and infinitive) and an adjunct clause (Phorm 799; causal clause). In continuous text
there is not an interchange of question and answer (although the author can sometimes pose questions and then
proceed to answer them himself). Nevertheless the sentences of a chunk of text can still be thought of as answering
implicit questions and thereby incrementally building

18
A. Who's that talking here? B. Sophrona (Phorm 739). A. But who are you looking for? B. Bacchis (Bacch 587). A. Whose daughter is he marrying? B. The daughter of our
neighbour here, the old man Euclio (Aul 289). A. What did he do? B. He silenced with his words the very angry old man (Phorm 477). A. What did you do? B. I sent him
down the stone quarries in fetters (Capt 944). A. What are you saying? B. That I saw his father, your uncle (Phorm 199). A. How come she can't be? B. Because they love each
other (Phorm 799).
14 INTRODUCTION

the information of the reader.19 Consequently they too display (at least) the binary informational structure just
illustrated for question and answer: except for out-of-the-blue utterances, part of the sentence will be new information
and part will be information that the author takes to be presupposed by the reader. If we found the following sentences
in a text
Bacchidem quaerito.
Filiam vicini huius Euclionis senis ducit.
Aio me huius patrem vidisse.
Non potest, quia uterque utriquest cordi.
we would analyze them in terms of a binary informational structure which, depending on the context, could
correspond exactly to the question-answer sequences on which they are modelled. Then in the first sentence, for
instance, the presupposition would be that I was looking for someone, and the new information would be that the
person in question was Bacchis. The sentence would mean ‘The person I am looking for is Bacchis,’ and not ‘What I
am doing to Bacchis is looking for her,’ nor ‘What's happening is that I'm looking for Bacchis.’
The basic binary division between new and presupposed information is not very subtle and comes nowhere near to
exhausting the range of pragmatic values we encounter in texts. There are different types of new information and
different types of presupposed information, and there are interesting ways in which new information can be treated as
old and old information as new. For the purposes of this work we shall adopt the following schema of pragmatic
values. There are three basic pragmatic categories: focus, topic and tail. Focus is prototypically new information, topic
and tail are prototypically old information. Focus is subcategorized as weak and strong focus. Strong focus in turn is
subcategorized as simple (exhaustive) strong focus, contrastive strong focus and counterassertive strong focus. Topic is
subcategorized as strong topic and weak topic. Topic and focus are not mutually exclusive, since strong topic can be
contrastively focused. Let's look more closely at this classification.
Weak focus is the category that has just been illustrated in question-answer sequences. As just noted, it also serves for
the straightforward communication of new information in continuous text. In this function it is sometimes called
informational focus. If you think of the hearer's information state as consisting of a database with a number of file
records or cards, then weak focus serves to fill in a blank field in one of those file records. So in the example from the
Bacchides we have been using, A knows of B that he is looking for someone (because he has just knocked on the door)
but he doesn't know who B is looking for; B's answer serves to fill the gap in this field of his database entry on B. As is
clear from the other examples, the scope (projection) of informational focus can vary. Outside question-answer
sequences with qu-words, the default scope for informational focus includes the verb
 INTRODUCTION 15

19
His Caesar imperat obsides quadraginta frumentumque exercitui … Illi imperata celeriter fecerunt (BG 5.20).
Illi (the Trinobantes) and imperata are old information; the nuclear verb phrase celeriter fecerunt is the weak focus: they
carried out Caesar's instructions and they did so quickly. (One can recognize a subarticulation to the extent that the
verb is more predictable than the adverb.) Where focus is confined to a single word, we call this ‘narrow focus,’ where
it projects beyond a single word we use the term ‘broad scope focus’; the latter term therefore covers a range of
different structures. Another type of weak focus is presentational focus
20
Erant apud Caesarem in equitum numero Allobroges ii fratres (BC 3.59).
Here the information being added to the reader's knowledge store is the presence (in a location previously established
in the context) of two Allobrogian brothers.
Simple strong focus is like weak focus in that it conveys new information to the hearer, but it does so in a less open-
ended way. If someone asks you who stabbed Caesar and you answer “Brutus,” your answer is not incorrect. It is true
that Brutus stabbed Caesar; it is also true that Cassius and a number of other conspirators did too, but you have
chosen to cite only a representative stabber or the most salient one in your answer. You feel that such an answer is
sufficiently informative: if the questioner had wanted an exhaustive answer, he could have forced one by asking “Who
all stabbed Caesar?”.20 On the other hand if you answer with a cleft (“It was Brutus that stabbed Caesar”), then you are
conveying that Brutus was the only one to stab Caesar (on that occasion), that Brutus exhausts the set of stabbers.
Weak focus may or may not be exhaustive, strong focus is interpreted as carrying an exclusive component in its
meaning. A contextually determined set of alternates to Brutus (the other conspirators or anyone else in the senate on
that occasion who did not like Caesar) is evoked. The assertion is then interpreted as holding for Brutus and negated
for the alternates. So weak focus tells us for Brutus whether he stabbed Caesar or not, strong focus tells us for all the
conspirators whether they stabbed Caesar or not. When strong focus is used to negate and correct a previous speaker's
assertion or a presumed assumption, it is called counterassertive focus
A. Mark Antony stabbed Caesar'. B. No! BRUTUS stabbed Caesar.
In contrastive focus the set of alternates is established on the basis of pragmatic rather than semantic factors. So you
can say
Not MARK ANTONY but BRUTUS stabbed Caesar.

19
Caesar demands of them forty hostages and grain for the army … They quickly carried out his orders (BG 5.20).
20
There were with Caesar among the cavalry two Allobrogian brothers (BC 3.59).
16 INTRODUCTION

The strong focus is exclusive relative to the contrasted alternate, not relative to the whole set of alternates.
The topic is informally defined as the entity about which the sentence is designed to convey information. (Paragraphs
or conversations can also have topics, sometimes called discourse topics or macrotopics, but we are not concerned
with those here.) In the file updating framework, the topic usually corresponds to the file record or card on which new
information is recorded. So if you are told
As for pizza, the cat won't eat it,
you can enter “not eaten by the cat” on the file card entitled “pizza.” Referential or generic expressions make good
topics: proper names (Brutus), definite descriptions (the leading conspirator) and pronouns (he) have readily identifiable
referents which are easily retrieved from the hearer's pre-existing knowledge (hearer-old information) or from anterior
stages of the discourse (discourse-old information) or from both. Nongeneric indefinite phrases frequently introduce
new referents into the discourse and consequently are difficult to topicalize
*As for a slice of pizza, it's under the table.
While weak topics are typically just items of old information of which new information is predicated, strong topics are
implicitly or explicitly contrastive
21
Fulvius in agrum Cumanum, Claudius in Lucanos abiit (Livy 25.19.6).
This illustrates that, as already remarked, topic and focus are not mutually exclusive: contrastive focus has been
superimposed on a topical constituent. While topic is principally associated with old information and focus with new
information, the correlation is not dependable. It is perfectly possible for old information to be (weakly or strongly)
focused whether it is topical or not, even pronouns
22
uter nostrum tandem, Labiene, popularis est, tune… an ego? (Pro Rab Perd 11).
It is also just about possible for even a dislocated topic to contain new information, as when a specific indefinite is
anchored to something definite
A dog on the other side of the street over there, it's chewing your bicycle seat.
Both topic and focus can also be recursive; that is, one can appear in a constituent embedded inside the scope of
another. In the pizza example above, As for pizza was a strong topic and the cat was a weak topic. (Subjects are default
weak topics in English.) The pizza is new information being introduced into

21
Fulvius departed into the territory of Cumae, Claudius into Lucania (Livy 25.19.6).
22
Which of us, I ask you, Labienus, is the true democrat, you or me? (Pro Rab Perd 11).
 INTRODUCTION 17

the discourse as a topic, the cat is old information already existing or reasily inferable from the discourse context.
Double foci are also common
Only Prof. JONES teaches VERSE composition.
Not all old information is topical; the sentence can include items of presupposed information without being designed
specifically to convey new information about those items
23
Movet feroci iuveni animum conploratio sororis…Stricto itaque gladio simul verbis increpans transfigit puellam.
(Livy 1.26.3).
In the second sentence Horatius' sister (puellam) is old information (cp. sororis) but she is not the topic of the sentence;
the sentence is not about her but about Horatius. Nor is she the focus; the sentence does not mean ‘the person that got
run through was the girl’ nor even ‘someone got run through and it was the girl.’ This type of old information is
sometimes called tail information (we shall adopt this term). Tails serve to lexically instantiate arguments that are
obligatorily projected but are not topics or foci, and at the same time to confirm the hearer's assumptions or refresh
his memory about old or inferable information. In some languages they often appear adjoined to the end of the clause,
where they can be thought of as antitopics. The partitioned structure of premodifier hyperbaton (see Chapter 6)
whereby the focused adjective stands at the beginning of the hyperbaton and the tail noun at the end is not
coincidental. The listener needs to attend more closely to the focus, while the presupposition is not part of what is
being asserted and tends to be taken for granted, as illustrated by the Moses illusion.21
It follows from what has just been said that the pragmatic structure of a sentence is largely determined by the context
(either via simple rules of correspondence or via an optimality calculus), but probably not entirely so. Some degree of
optionality seems to remain; the same applies to the application of the syntactic rules used to encode the pragmatic
structure.

Notation for semantic and pragmatic meaning

There are obvious reasons why the study of algebra and geometry is not conducted entirely in prose, however lucid
and elegant, and linguists have found that algebraic aspects of semantics and geometrical aspects of syntax are also best
expressed formally. Formalism adds clarity and precision (at the cost of the loss of some nuances). We will briefly
characterize those aspects of the formalism that are associated with concepts relevant for our study of word order.
This section deals with the semantics, the next section with the syntax. To keep things simple, we will use English
examples in this section.

23
The lamentation of his sister angered the fierce young man. So drawing his sword, while shouting reproaches at her, he ran it through the girl's body (Livy 1.26.3).
18 INTRODUCTION

The arguments of a predicate are written in parenthesis after the predicate

Run (j): ‘Jack runs’
Student (j): ‘Jack is a student’
Stab (b, c): ‘Brutus stabs Caesar’
Brother (q, m): ‘Quintus is the brother of Marcus’
Fond (j, s): ‘Jack is fond of Statius.’
If the predicate has more than one argument, the conventional order of the arguments is that of the grammatical
relations in English, therefore subject before object. Consequently Stab (b, c) means that there is a stabbing relation
between Brutus and Caesar such that Brutus is the stabber and Caesar is the stabbee. The default order of semantic
composition is object before subject: this can be explicitly formalized by using the socalled curried notation: (Stab (c))
(b). This expression means that Brutus has the property of stabbing Caesar. The predicate stab applies to its internal
argument Caesar to form a complex function stab-Caesar, which in turn applies to Brutus. Predication is interpreted as
set membership: so Caesar is a member of the set of those who got stabbed and Brutus is a member of the set of those
who stabbed Caesar. Since grammatical relations themselves have a default correlation with pragmatic values, this
notation is more suitable in pragmatically sensitive work. However for the sake of simplicity we shall use the uncurried
format and assume automatic translation into the curried format.
There are two types of keys on a keyboard: the alphanumeric keys have a straightforward “denotational” function,
while other keys such as caps, delete, change font, etc. serve to perform operations on alphanumeric characters. The
same sort of division of labour is found in semantics and syntax. The major lexical categories are basically
denotational, but quantifiers, negatives, conjunctions, tense markers, interrogatives and focus have a more logical
function, and so they are often called operators.
If instead of Student (j) we write Student (x), where x is a variable, we mean that whoever the variable x is assigned to
is a student. If Student (x) is a complete sentence, then the variable is a free variable, and its assignment will be
recovered by the normal process of anaphora resolution in discourse; for instance ‘Jacki came into the room. Hei is a
student.’ The referent of the anaphoric pronoun he is indicated by the subscript ‘i’; this is called coindexation. Variables
can also be bound by an operator. The lambda operator is used to define a set in terms of its characteristic function
(rather than by listing its members): λx. Student (x) means the set of all x such that x is a student. If we substitute Jack
for x in this expression, we get the proposition that Jack is a member of the set of students, i.e. Student (j) as before.
The lambda operator is particularly useful for the construction of complex properties
λx. Student (x) ∧ Like (x, s): ‘students who like Statius.’
Variables are also used to express quantification. The simplest formalism is that used in predicate logic
 INTRODUCTION 19

∃x. Gaul(x) ∧ Kill (Caesar, x): ‘Caesar killed a Gaul’

∀x. Gaul(x) → Kill (Caesar, x): ‘Caesar killed all the Gauls.’
The formulae say, respectively, ‘There is someone who was a Gaul and who got killed by Caesar,’ and ‘For all persons,
if they were Gauls, they got killed by Caesar.’ There are various problems with this formalism; the one that concerns us
here relates to the pragmatics. Consider the following
¬∃x. Student (x) ∧ Like (x, Statius).
This formula has various readings. One (the socalled weak22 reading) simply denies the existence of anyone who is
both a student and a fan of Statius. There are also two socalled strong readings, which probably involve an existential
presupposition: ‘No student (none of the students) likes Statius,’ and ‘No fan of Statius is a student.’ Both strong
readings are available, since nondynamic conjunction is commutative. However, even if the two strong readings are
truth-conditionally equivalent, they carry different presuppositions, they may differ in the procedures that would be
used to verify them, and they are pragmatically different. One tells us something about students, the other something
about fans of Statius. So for strong quantification we shall use what is called restricted quantification, which uses a two-
place operator (instead of the one-place operator of standard predicate logic). This affords an explicit articulation into
pragmatically correlated components
∃x | Student (x) | Like (x, Statius): ‘A student likes Statius’
∀x | Gaul (x) | Kill (Caesar, x): ‘Caesar killed all the Gauls.’
The second example says that for every assignment of the variable x to a Gaul (present in the situation under
discussion), x got killed by Caesar. This type of structure is known as a tripartite structure; we have used vertical bars
to separate the three fields. The first field contains the operator ∀ which binds the variable x. The second field is the
restrictor: it restricts the set of those killed to those who are Gauls. The third field is called the nuclear scope. The
second and third field have default pragmatic correlates. The nuclear scope is typically new information (focus). The
restrictor is typically presupposed or accommodated information: material directly available from the sentence is
supplemented and calibrated with contextual information to define the domain of quantification. Of course, these
default pragmatic correlates can be modified in pragmatically marked sentences (e.g. CAESAR killed all the Gauls, with
narrow subject focus and presuppositional verb phrase).

Focus
Focus is akin to quantification in that it entails ranging over the set of alternates and picking that one which satisfies the
presupposition. So given the pragmatic articulation of the tripartite structure just noted, it is a natural move to analyze
focus in terms of the tripartite structure, using a silent operator FOC in place of the overt quantifier in quantificational
sentences. This gives a convenient
20 INTRODUCTION

and succinct semiformal representation, which we will use most of the time. Suppose the question under discussion is
who Caesar killed; then answers like ‘Caesar killed (some) Gauls,’ or ‘Caesar killed Dumnorix’ can be expressed in
tripartite structures such as the following
FOCx | Kill (Caesar, x) | Gaul (x) ‘Caesar killed GAULS’
FOCx | Kill (Caesar, x) | x = Dumnorix ‘Caesar killed DUMNORIX’
FOCP | ∃x.Kill (Caesar, x) ∧ P(x) | P = Gaul ‘Caesar killed some GAULS.’
Here the restrictor clause is the presupposition and the nuclear scope is the focus. The tripartite structure is very clear
about the articulation of the sentence into focus and presupposition, but it is less clear about exactly how the focus is
composed with the presupposition to get an interpretation.23 In fact there are a number of different semantic scenarios,
which we shall now proceed to review. The main reason why we chose to use tripartite structures is that they provide a
pragmatically partitioned meaning without the added complications of an explicit compositional semantics. They are
also a convenient one-size-fits-all formalism that does not force a choice among the various possible subtypes of focus
meanings. So if you are not interested in the technical details, you can skip this part of the discussion.
The first and simplest way to interpret tripartite focus structures is like restricted first order quantifiers: this is the
formula view, which we will use frequently in this book. The restrictor and the nuclear scope each contain an open
formula with a free variable (which then gets bound by the operator). So each field is an open proposition: ‘x got killed
by Caesar,’ ‘x is a Gaul.’ The nuclear scope can be a simple predication as in the first example (Gaul(x)) or it can be a
specification (equation) as in the second example (x = Dumnorix). The last example extends the coverage of the
formula view, in that FOC is a higher order operator ranging over properties, and the restrictor contains an explicit
existential presupposition (‘Caesar killed some people having property P’). This contrasts with the other examples,
where arguably the killing of someone by Caesar is a matter at issue rather than a presupposition. Since the formula
view mimics first order quantification, it is open to the same sort of criticism. It assigns a consistent biclausal structure
to all focus, whereas language can draw a distinction between clefts and monoclausal types of focus.
In the second interpretation of the tripartite structure, the predicate view, the two open formulae of the first
interpretation are translated into expressions denoting sets
FOC | λx.Kill (Caesar, x) | λy.Gaul (y)
FOC | λx.Kill (Caesar, x) | λy.y = Dumnorix.
Now the focus operator can be interpreted as a second order quantifier, giving the familiar relational (uncurried)
generalized quantifier semantics. The variables are not bound by FOC but by the lambda operator. In the case of
strong, exclusive focus, FOC would have approximately the same semantics as only.24
 INTRODUCTION 21

ONLY (A, B) means B ⊆ A ‘B is a subset of A,’ or equivalently A ⊇ B ‘A is a superset of B.’ Only Dumnorix is the set of
sets having only one member to which Dumnorix belongs, i.e. the set of Dumnorix's singleton properties. But if we
want to preserve the regular correlation of restriction with presupposition (and nuclear scope with focus), we should
first compose FOC with the restriction and not with the nuclear scope.25 So we can appeal to the fact that only
functions as the converse of all, to get the reading ‘All those killed by Caesar were members of the set of individuals
equal to Dumnorix.’ Apart from anything else, this is probably too strong for nonexclusive weak focus sentences,
which look more like a restructured version of ordinary predication.26
Such a restructuring of predication is achieved in the third interpretation, the socalled structured meaning theory.
Structured meanings are derived by lambda abstraction over a variable of the same type as the focus. In the case of
Dumnorix, we abstract over an individual
<λx. Killed (Caesar, x), Dumnorix>.
The expression before the comma is the restriction (‘the set of those killed by Caesar’) and the expression following the
comma is the focus. The normal assumption is that the lambda abstract is applied to the focus to get the interpretation
of the whole sentence: ‘Dumnorix is a member of the set of those killed by Caesar.’ This correlates the focus (F, i.e.
Dumnorix) with the argument and the background (B) with the function, which is a reversal of normal subject-
predicate relations. Moreover, as already noted, strong focus at least (only Dumnorix) seems to be intrinsically
quantificational.27 These problems can be fixed by shifting the types28 of one or both the expressions in the structured
meaning.
In the fourth interpretation of the tripartite structure (typeraised structured meaning) the second expression of the
structured meaning is typeraised to become a functor (type <et,t> in our example) taking the first expression as its
argument
<λx. Killed (Caesar, x), λP.P (Dumnorix)>.
This additional step reverses the function-argument relation of the structured meaning, so that the result is not B(F)
but F(B), giving the interpretation ‘Getting killed by Caesar is a property of Dumnorix.’
In the fifth and last interpretation (typelowered structured meaning), Dumnorix retains the type of an individual and
the type of the restrictor is lowered from set to individual (<et> to <e>). Since the two expressions of the structured
meaning are now of the same type, composition takes the form of an equation
ιx. Killed (Caesar, x) = Dumnorix.
Here the iota operator is used to create a definite expression (‘the person who Caesar killed’), which is equated with the
focused proper name. The result is a cleft-like semantics that is appropriate for strong narrow focus: ‘the one that
22 INTRODUCTION

got killed by Caesar was Dumnorix.’ The iota expression is potentially a description for any one of the whole set of
alternates; each alternate is a victim of Caesar in a different set of worlds, and the focus serves to pick out that alternate
which it describes in the actual world.29 The equative analysis is supported by the typological observation that clefts can
use an equative rather than a predicative copula and that focus markers often develop diachronically from cleft
constructions.30
The above semantics for focus applies whether the language normally separates the focus from the presupposition
(cofocus) in the syntax (like Latin) or normally keeps the focus in situ (like English). That there must be some sort of
separation of focus and presupposition, at least in the semantics, in English too is clear from sentences with
restructuring due to subject focus
Seven/SEVEN students paid $35 for a taxi to the game
Prof. Jones/JONES likes his students, and Prof. SMITH does too
Many Stanford/STANFORD students became astronauts.
The first sentence can have a collective or distributive reading. On the collective reading, the students paid $5 each for
a taxi ride they took together; on the distributive reading they paid $35 each for a longer ride each one took
individually. Strong focus on the cardinal tends to induce the distributive reading by creating a cofocus that is a
property characterizing each of n students. The students then have to be part of the presupposition and cannot be part
of the focus; the strong focus sentence cannot mean ‘There were seven students and they paid $35.’ This precludes
simple reference to the plurality as a collective entity and forces universal quantification over its individual members.
The same restriction applies to numerals that are overtly separated from their noun by postmodifier hyperbaton in
Japanese.31 In the second sentence the pronoun his is part of the ellipsis in the second conjunct, where it can have a
strict or sloppy reading.32 The strict reading means that Prof. Smith likes Prof. Jones' students, and the sloppy reading
means that Prof. Smith likes Prof. Smith's students. Strong focus on the names of the professors tends to induce the
sloppy reading by creating a cofocus ‘x likes x's students' which is a property that applies to each professor. The last
sentence could mean that a significant proportion of Stanford students became astronauts or (more sensibly) that a
higher proportion of Stanford students than students from other universities became astronauts. Focus induces the
second reading which, arguably, results from the creation of a cofocus ‘many students from P university became
astronauts.’33 It would be possible to massage English syntax at a postsurface level (logical form) to express pragmatic
structure more systematically, as in the movement theory of English focus.34 On this approach, not only is there no
variation from one language to another in the semantics of focus, there is also no variation in the syntax of focus: focus
movement applies either overtly (in the surface syntax) or covertly (at logical form). But that could be considered an
unwarranted intrusion of semantics into syntax. If so, then prosodic structure
 INTRODUCTION 23

is mostly translated directly into pragmatic structure in English, without the mediation of syntax. These reservations do
not apply to free word order languages like Latin, since phrases actually show up in pragmatically dedicated positions:
we don't have to move them covertly because, more often than not, they are already where we want them overtly.
Languages like Latin present us with the converse problem: is grammatical meaning translated directly from the
inflectional endings, or are the inflected argument phrases moved back into their appropriate underlying structural
positions for semantic interpretation?35 When inflectional distinctions are neutralized, word order rules can become
more rigid (freezing under morphological ambiguity),36 as for instance with subject and object phrases in the accusative
and infinitive construction. But exceptions still occur
24
magnum tamen exercitum Pompeium habere constat (Ad Fam 6.18.2).
Such exceptions can often be resolved by prosodic information and by animacy distinctions.
So there are three possible perspectives on “free” word order. We started with the familiar observation that grosso
modo English word order is fixed and Latin word order is free. We then introduced the refinement that, again very
roughly speaking, Latin word order is grammatically free but pragmatically fixed, while English word order is
pragmatically free but grammatically fixed. Finally we entertained the idea that the grammatical freedom of Latin word
order and the pragmatic freedom of English word order could be a superficial consequence of the fact that where
grammatical and pragmatic order diverge, it is not possible to express both at the same time in a single surface order;
but that this problem could easily be rectified by covert movement at logical form. In that sense (perhaps with some
exaggeration), grammatical and pragmatic word order would both be fixed in both languages. One recurrent problem
that remains is deciding whether movement is purely syntactic, purely prosodic, or is syntactic movement for prosodic
reasons. Even well-known cases of syntactically defined focus positions, like Hungarian focus, have recently been
reanalyzed in prosodic terms.37 In Classical linguistics, the idea that the prosody of focus is responsible for some word
order effects has a long history. Weil writes: “le changement des accents entraîne d'ordinaire un changement de l'ordre
des mots.”38 In its strongest form, the prosodic theory entails that properly syntactic structure is “erased” (invisible)
and purely prosodic constituents move to the edges of superordinate prosodic constituents.

Type theory
One way of defining words and larger grammatical constituents is in terms of their combinatory properties. For
instance a transitive verb can be defined as a functor looking for a direct object noun phrase as its argument to make a
verb

24
It is known however that Pompeius has a large army (Ad Fam 6.18.2).
24 INTRODUCTION

phrase, in the notation of categorial grammar VP/NP. Then the string VP/NP NP → VP by functional application;
the NPs cancel out as in fraction multiplication (½x2=1), yielding the superordinate category VP as the result. The
verb phrase in its turn is a function looking for a subject to make a sentence: NP S\NP → S. In tandem with this
syntactic operation there is a corresponding semantic operation that combines the meanings of the verb and the object
into the meaning of the verb phrase. The combining semantic elements can likewise be defined in terms of their
combinatory properties in what is known as the theory of types. There are two basic types, e for ‘entity,’ the denotation
of referential noun phrases and t for ‘truth value,’ the denotation of sentences. Other syntactic categories are defined in
terms of combinations of these two basic types. For instance a verb phrase has the type <e,t>, a function from an
entity (a subject noun phrase of type <e>) to a truth value (the sentence of type <t>). When <e,t> applies to <e>, the
two <e>'s cancel out as before: <e> <e,t> → <t>. Adverbs that are verb phrase modifiers can be assigned the type
<<e,t><e,t>>, that is functions from an expression of type <e,t> into an expression of type <e,t>; and so on. This
way of looking at the semantics of syntactic categories is particularly useful when a single prima facie syntactic category
corresponds to more than one semantic type. In Latin nouns can stand for definite noun phrases, specific indefinite
noun phrases, nonspecific indefinite noun phrases, and predicates. Adjectives can stand for predicates, attributive
modifiers and noun phrases. The necessary distinctions can often be made in terms of type theory. Even in English,
which has overt determiners, type theory is useful for distinguishing different determiner phrase meanings
The Gallic chief attacked the Romans
A Gallic chief (his name was Indutiomarus) attacked the Romans
Some Gallic warriors attacked the Romans, the rest stayed behind
Jack was a student
Jack had a beer.
The first two examples have a definite and a specific indefinite subect phrase, respectively; these can be assigned the
type <e>. In the third sentence the subect phrase is a strong (partitive) quantifier, assigned the type <et,t>. It is
sometimes helpful to be able to interpret a noun phrase in a higher type. For instance, a subject noun phrase of type
<e> can be interpreted as a quantifier of type <et,t>; this is called typeraising or lifting. In the penultimate example the
indefinite noun phrase a student is a predicate, therefore of type <e,t>. In the last example a beer is a nonspecific
indefinite, arguably also of type <e,t>. Some of these distinctions are more overtly encoded in languages like
Norwegian, Albanian, Turkish and Persian. Type theory is probably the appropriate mechanism for representing
differences of informational individuation, which, as we shall see, underlie a number of rules for Latin word order.
 INTRODUCTION 25

Syntactic structure
We should make it clear at the outset that this book is not about Latin syntax narrowly defined as an object of study for
its own sake, but about how Latin syntax functions as a vehicle for pragmatic and semantic meaning. We need syntax
because it is the interface between word order and meaning, but we will be adequately served by a fairly simple,
concrete and, for the most part, traditional syntactic framework. Consider the following from the first paragraph of
the first book of the Gallic War
25
pertinent ad inferiorem partem fluminis Rheni (BG 1.1).
Despite the comparative freedom of Latin word order this is not just a string of independent words that could have
been arranged in any order: *ad pertinent Rheni for instance is not allowed, and other orders which are allowed would be
considered less natural, particularly in prose. Most people would be willing to accept that this example includes the
following constituents: a noun phrase (NP) fluminis Rheni; a noun phrase inferiorem partem fluminis Rheni; a prepositional
phrase (PP) ad inferiorem partem fluminis Rheni; and a verb phrase (VP) pertinent ad inferiorem partem fluminis Rheni. The
appositional noun phrase fluminis Rheni is the complement of the relational noun partem; the whole noun phrase
inferiorem partem fluminis Rheni is the complement (object) of the preposition ad; and the prepositional phrase ad inferiorem
partem fluminis Rheni is the complement of the verb pertinent. In each case a phrase XP is made up of a head after which
it is named (noted X or X°), and its complement YP. Each expansion of the head X with an argument YP is called a
projection of X. Since we are not going to assume an empty determiner position corresponding to English the for
Latin, the phrase partem fluminis Rheni is a complete noun phrase (XP). It is further modified by the adjective inferiorem.
The category and attachment of the adjective are not entirely clear (see Chapter 5); for the sake of simplicity, we have
treated it as adjoined to XP, that is extending XP into a superordinate XP. The resulting hierarchical structure, called a
configuration, is depicted in the tree diagram in Figure 1. The lines are called branches; the ends of the branches are
called nodes, and each node is marked with a grammatical label. Nodes whose branches join at the immediately higher
node (the mother node) are called sisters. Each complement is the sister of its head. The appositional phrase is
represented by a triangle since it is not further analyzed.39
One of the characteristic features of Latin syntax is that it has pragmatically defined functional projections
superordinate to XP which are crosscategorial. We define these as FocXP and TopXP. FocXP is a focus position local
to the phrase XP, and TopXP is a topic (subject) position local to the phrase XP. We sometimes refer to these two
positions as specifier positions. (Traditional

25
They reach the lower part of the Rhine river (BG 1.1).
26 INTRODUCTION

Figure 1: Example of tree structure Pertinent ad inferiorem partem fluminis Rheni (BG 1.1)
Spec VP is decomposed into two functional projections, TopVP and FocVP, neither of which hosts the subject of the
sentence.) A complement of X can therefore appear either in the basic complement position (sister of X) or in either of
the two higher functional projections, as illustrated in Figure 2. The three potential YP-positions form what is known
as a chain. YP has access to multiple positions in the tree and is attracted to whichever position is appropriate to its
pragmatic value. The XP-tree is split into two layers, the basic projection of X (XP) and a superstructure consisting of
its extending functional projections (FocXP and TopXP); each layer is filled on the basis of the pragmatic value of the
candidate phrase(s). The tree configuration is determined primarily by discourse properties like topic and focus rather
than by grammatical properties like subject and object. Consequently Latin is called a discourse configurational
language.40 The pragmatically defined positions are taken to be specifiers in a full phrasal projection, rather than merely
adjoined phrases. The difference is that the former have an empty head position (not represented in Figure 2), the
latter do not. English does not have this crosscategorial system of pragmatically defined functional projections, but it
does have a system of noun phrase (determiner) and verb phrase (auxiliary) operators that serve the semantic purpose
of linking these constituents with their discourse referents, individuals and events respectively. In Latin these operators
are not lexicalized and arguably do not project syntactically. The focus position is higher and to the
 INTRODUCTION 27

Figure 2: Functional projections in a nonderivational theory

left of the complement position in the tree: this is an example of the structural relation known as c-command, which is
often the syntactic correlate of the semantic relation of scope. (The scope of a variable binding operator is the domain
within which it binds a variable.) The discourse functional projections do not have lexicalized heads but project
positions into which arguments and heads can move.
As already pointed out, the tree-splitting of XP is crosscategorial: it applies more or less equally to noun phrases, verb
phrases and adjective phrases, and elements of it are recognizable even in prepositional phrases. Higher layers of
functional projections are recognizable at the clausal level. The lowest layer is the VP layer; it contains the nuclear
assertion. The intermediate layer contains the subject and phrases scrambled out of the VP; we will call this layer the
IP layer (without entailing any particular assumptions about the syntactic relation of subjects to verbal inflection or
auxiliaries; we do not posit an inflection or tense phrase separate from the verb phrase).41 In gross terms, the VP layer
hosts a bare description of the event42 in terms of the new information it contains, while the IP layer hosts material that
anchors the event to the discourse context. The highest layer contains a collection of operators: complementizers
(subordinating conjunctions), strong topics, interrogatives and some foci. We will call this layer the CP layer after the
complementizers.43 Figure 3 is designed to give a general illustration of this layering without exhausting the possibilities
and without insisting on a distinction between phrasal projection and adjunction. It is likely that the potential for higher
levels of functional projection is crosscategorial too, since it is probably discernible in more complex noun phrases.
The existence of pragmatically defined crosscategorial functional projections is the structural actuation of discourse
configurationality.
28 INTRODUCTION

Figure 3: Layers in Latin clausal structure

The functional projections serve to identify the pragmatic values of the expressions they host, and the order in which
they are presented to the listener presumably reflects the order in which, typically, this information would be most
naturally processed by the listener and interpreted in its appropriate hierarchical structure.
Although we do not actually believe that YP starts out at the foot of the chain (its basic complement position in XP)
and then moves to a functional position higher in the tree, the movement metaphor provides such a convenient
framework in which to discuss Latin syntax that we shall adopt it. An example of this type of analysis is given in Figure
4, which depicts a focused complement YP of X as having moved from the complement position (sister of X) to the
specifier of FocXP. A derivational perspective brings with it some
 INTRODUCTION 29

Figure 4: Focused complement in a derivational theory

convenient terminology. We will speak of YP raising to a higher position, using the term raising in its most general
sense.44 According to the schema set out above, movement is exclusively (or almost exclusively) from a lower right
position to a higher left position. However we shall occasionally mention extraposition analyses, according to which an
element moves to a higher position to the right. When a phrase raises, it can leave behind some part of itself in its base
position; this is called stranding. For instance stranding is a possible basis for hyperbaton in a derivational perspective:
cepit magnam urbem → magnam cepit urbem, where the adjective raises and the noun is stranded. Note that the movement
theory creates a complication in the analysis of hyperbaton: the constituents may need to be able to change their
syntactic category and semantic type when they become discontinuous (see some of the trees in Chapter 6). Conversely
if a phrase takes with it some element that it should have left behind when it raises, that is called piedpiping: cepit quam
urbem? → quam urbem cepit?, where the interrogative piedpipes the noun although that is not part of the query but part
of the presupposition. Piedpiping allows focus movement to respect the integrity of the syntactic phrase, just as it
respects the integrity of the word
CAECILIUS I like, not LUCILIUS
*CAE I like -cilius, not LU-cilius.
If an element moves without changing the serial order of the words, that is called string vacuous movement: In Quis
cepit urbem?, one can posit that the subject pronoun raises from a subject position to an interrogative position without
changing the order of the words.

Prosodic structure
We add a few very brief remarks on prosodic structure, since at various points in the text we recognize the possibility
that word order perturbations are triggered by prosodic requirements (without making a serious attempt to work out
30 INTRODUCTION

a coherent theory based on a wide range of data). We tentatively recognize three levels of prosodic structure: the word
or clitic group (ω), the minor phrase (φ) and the intermediate or major phrase (Φ). We are assuming that Latin minor
phonological phrases typically consist of two words and are trochaic, that is the leftmost word is prominent; one word
minor phrases can also occur, and three word minor phrases with hierarchical structure. Then using the grid formalism
we can represent the prosodic structure of a popular septenarius45 as follows
x x (Φ)
x x x x (φ)
x x x x x x x (ω)
[ [Postquam Crassus]φ [carbo factus] φ]Φ [ [Carbo crassus]φ [factus est]φ]Φ.

In view of the chiasmus, the phrase after the diaeresis may have had stress inversion or two equally salient stresses, but
we have glossed over this complication. The top layer of x symbols indicates stress in the intermediate or major phrase
(Φ), the second layer in the minor phonological phrase (φ), and the lower layer represents word stress (ω). According
to this approach, the prosodic structure can trigger word order movements when there is a clash between
pragmatically induced stressing and the default stress patterns assigned by the prosodic structure. For instance if a
minor phrase consisted of a weakly stressed tail word followed by a strongly stressed focus word, there would be a
clash between the iambic W(eak) S(trong) structure induced by the pragmatics and the trochaic SW structure required
by the overall rhythm of the language. This clash would be eliminated by inverting the order of the two words either in
the syntax (assuming it can look ahead to the phonology) or in the phonology (after the syntax is over and done with).

Postscript
This is the end of our overview of the theoretical concepts and related formalism used in this study of Latin word
order. While it is fairly complete, for obvious reasons it is also very compact, and readers unfamiliar with the subject
may find it helpful to look at some (English-oriented) introductory texts which cover the same sort of material in
greater detail and at a more leisurely pace. We list a few in the bibliography below. In the following chapters, we have
deliberately placed most of the technical discussion (not quite all of it) in dedicated sections entitled “Structural
analysis.” Linguistic particulars and specific bibliographical references are further relegated to the endnotes. The
resulting three-tiered structure gives readers some freedom to choose the level of linguistic detail that best fits their
own interests. Readers who so wished could conceivably skip the structural analysis sections and the endnotes
altogether and still use the rest of the book as a largely descriptive account of Latin word order in the simple sentence.
While it is possible to use the book in this way, it is
 INTRODUCTION 31

hardly advisable. Twentieth-century theoretical linguistics is not, as Blackadder said of the Renaissance, just something
that happened to somebody else. By giving us access to a previously unattainable, in fact unimaginable, depth and
precision of analysis, it has profoundly affected our understanding of every aspect of natural language, and there is no
reason to think that Latin word order should be exceptional in this regard.

BIBLIOGRAPHY
Cann (1993); Carnie (2002); de Swart (1998); Heim & Kratzer (1998); Kadmon (2001); Lambrecht (1994); Ouhalla
(1999); Roberts (1997); Tallerman (1998).
1. The instructor means ‘at a meeting of the senate in the Theatre of Pompey’: in… senatu… in curia Pompeia (De Div
2.23).
2. We use the term “pragmatic” in this book in the narrowly defined sense of ‘pertaining to informational structure’
(Vallduví 1992; Hendriks 2002). It is familiar in this sense, and less cumbersome than “informational.” The
distinction between semantics and pragmatics broadly defined is a complicated philosophical and linguistic
question and currently a hotly debated issue (e.g. Turner 1999).
3. Inspired perhaps by Henri Weil's examples of topicality: Idem ille Romulus Romam condidit; Hanc urbem condidit
Romulus; Condidit Romam Romulus: “la syntaxe [i.e. grammatical relations, not constituent structure] est la même
dans ces trois phrases… Pourtant on dit dans les trois phrases des choses différentes” (Weil 1869:24).
4. This does not always follow, since pragmatic distinctions sometimes do have truth-conditional consequences, as
noted later.
5. While free word order can be exploited for rhetorical and artistic purposes, it is not created for those purposes,
but is an intrinsic property of the syntax of the language.
6. The reasons for its neglect have more to do with the accidents of academic priorities than with the realities of
intellectual progress. The philological discoveries and insights of the late nineteenth and early twentieth centuries
never made it into the textbooks, a complaint already voiced by Linde (1923). On the theoretical side, the
interface of syntax with pragmatics has not been high on the list of programmatic needs for many linguists until
very recently (Hulk & Pollock 2001; Bailyn 2001). This attitude can be traced back to the pioneering work of Ross
([1967]/1986), who tentatively attributed free word order to a low level stylistic reordering of constituents, and
lives on in the notion that word order variation belongs in a postsyntactic “phonological” component. The
distinction between postsyntactic argument linearization and phonologically driven syntactic movement is studied
by van Gelderen (2003). It is interesting that the same perspective was prevalent at the beginning of the twentieth
century: “Die Fragen der Wortstellung sind lange als eine Art Kuriosum behandelt und in die Rumpelkammer der
Stilistik verwiesen worden” (Kroll 1920); “The rules of word-position have too long been the Cinderella of
linguistic science” (Jespersen cited by Ullman 1919).
7. A few illustrations of manuscript variation from the sixth book of the Gallic War are cited by Hering (1987):ex
reliquis hostes partibus – hostes ex reliquis partibus (BG 6.37)
32 INTRODUCTION

totis trepidatur castris – trepidatur totis castris (BG 6.37)quanto res sit – quanto sit res (BG 6.38)deiecti se in
signa – deiecti in signa se (BG 6.40).For more examples see Ullman (1919). In his edition of the Pro Cluentio,
Peterson (1899) says that there are over a hundred instances of “meaningless” word order variation between the
two manuscript traditions. There is a dissertation devoted to this phenomenon in Cicero (Rönsch 1914), which
includes many instances of word order variation in the structures analyzed in this book. Here are some
examples:VERB INITIALDicit accusator haec / Accusator dicit haec (Pro Clu 81)COPULA, AUX, EXIST RAISINGSed
pleni sunt omnes libri / Sed pleni omnes sunt libri (Pro Arch 14)cum hospes esset / cum esset hospes Heiorum
(Verr 2.4.6)Omnes in illo sunt / Omnes sunt in illo rege virtutes (Pro Reg Deiot 26)Cn. Magius est mortuus /
mortuus est (Pro Clu 21)WEAK PRONOUNin quot se laqueos / laqueos se induerit (Verr 2.2.102)in foro sibi
medio / in medio foro sibi (Verr 2.3.105)GENITIVEad Siciliae civitates / ad civitates Siciliae (Ver 2.3.44)provinciae
spolia / spolia provinciae (Verr 2.5.59)patris lacrimae / lacrimae patris (Verr 2.5.109)ADJECTIVEa viro improbo /
ab improbo viro (Pro Clu 189)muliebrem libidinem / libidinem muliebrem (Pro Cael 1)clara voce / voce clara
(Pro Caec 22)QUANTIFIERlocis omnibus / omnibus locis (Verr 2.2.154)aratores omnes / omnes aratores (Verr
2.3.112)MODIFIER HYPERBATONpaternas haberet arationes / paternas arationes haberet (Verr 2.3.97)a duobus
potentissimis regibus infertur / adfertur regibus (Pro Leg Man 4)Quem enim imperatorem possumus /
possumus imperatorem (Pro Leg Man 37)fructum caperes maiorem / fructum maiorem caperes (Pro Sull
90)CONJUNCT HYPERBATONmultitudinem criminum et atrocitatem / multitudinem et atrocitatem criminum (Pro
Clu 81)clarissimi viri atque amplissimi / clarissimi atque amplissimi viri (Pro Clu 95).
8. In practice our attention will mainly be directed to questions (ii)–(v).
9. Jacobs (2001).
10. The process by which this type of structure is derived is sometimes called Y-movement. Depending on various
factors, there may or may not be an intonational break after the topic.
 INTRODUCTION 33

11. We intend structured meaning here to be understood quite generally, not restricted to the technical sense of the
expression (discussed below) associated with the work of Cresswell and Jacobs among others. This generalization
reflects the homomorphism of syntactic structure and pragmatic meaning in Latin.
12. According to a narrow definition of syntax, pragmatic features are entirely extragrammatical; the role of grammar
is just to establish which word orders can be generated by which projections and which movements. But in
discourse configurational languages like Latin, the syntax contains not only projections that correlate with
semantic categories, but also projections that correlate with pragmatic categories. Since information structure has
been grammaticalized (Jelinek 2000), the relevant generalizations cannot be expressed in a noncircular way if the
latter are excluded. At the other end of the spectrum lies an extreme version of the functionalist approach to
word order which dispenses with (or is noncommittal about) syntactic structure. But just as discourse
configurational languages cannot be analyzed without access to discourse, so they cannot be analyzed without
configurations. To find out how linear order translates into pragmatic meaning, we need to establish three things:
the syntactic structure, the discourse structure, and the interface between them. The point is not that ecumenical
tolerance is preferable to divisive factionalism, but just that “pragmatics without syntax is empty; syntax without
pragmatics is blind” (Y. Huang). Newmeyer (1998) compares formalist and functionalist approaches to language.
13. Some of the category labels are unorthodox, but these could easily be changed.
14. Other research strategies that have been used include statistical analysis of categories or individual lexical items,
and the collection of particularly telling examples in suport of a thesis. The former strategy is mostly too blunt to
be of much practical use, although it does provide a useful preliminary orientation. The latter has yielded many
valuable insights, but the evidence presented may sometimes be insufficiently balanced and comprehensive.
15. Tesnière (1959); Fillmore (1995).
16. Bartsch (1987).
17. Bresnan (2001).
18. Happ (1976) is a study of valency in Latin.
19. Van Kuppevelt (1995); Roberts (1996); Martí (2003).
20. For exhaustive listing questions in Kashmiri the interrogative is reduplicated: kyaa ‘what?’, kyaa-kyaa ‘what all?’
(Bhatt 1999).
21. The answer to the question ‘How many animals did Moses take on the ark?’ is not ‘Two of each’ but ‘None’; it
was Noah who took animals on the ark. Here is another example from the literature: ‘When the plane crashed,
where were the survivors buried?’.
22. In a generalized quantifier perspective, weak quantifiers like some are symmetrical, which means that their
arguments can be inverted without affecting truth: D(A)(B) = D(B)(A). Strong quantifiers like most or all are not
symmetrical: D(A)(B) ≠ D(B)(A). All students like Statius is not the same thing as All fans of Statius are students,
unless it also happens to be the case that only students like Statius. The situation is complicated by the fact that
weak quantifiers can sometimes be used partitively, so that they get a strong reading: Some (of the) students like
Statius.
23. Hajičovà, Partee & Sgall (1998). Heim (1997) discusses the merits of predicates versus formulae in
quantificational structures; see also Sauerland (1998).
24. Not all strong foci can be paraphrased with only: (*Only) ALL the students passed the test; (*Only) MARY arrived first.
(Atlas 1996; Kiss 1998; 2001).
34 INTRODUCTION

25. Only (plus NP) is not a regular determiner type quantifier but the mirror image of one. It is conservative on its
right argument rather than on its left argument, and its left argument rather than its right argument hosts the
focus. In Some students smoked, students is the presupposition, but in Only Classics students smoked, Classics students is or
includes the focus (de Hoop 1995; von Fintel 1997).
26. Weak focus sentences and strong focus sentences have different word order in Finnish (Vallduví & Vilkuna 1998)
and Hungarian (Kiss 1998).
27. Van der Linden (1991).
28. See the next subsection on type theory.
29. This can be expressed overtly by intensionalizing the semantics; for instance in the structured meaning format
<λxλw. Killed (w) (Caesar, x), Dumnorix>.
30. Ouhalla (1999a); Heine & Reh (1984).
31. Nakanishi (2003). With “once-only” predicates a collective reading is unavailable and a distributive reading is
excluded, so hyperbaton is ungrammatical: Men yesterday three John's mother hit, *Men yesterday three John's mother killed
‘Three men hit/*killed John's mother yesterday.’ In Turkish, Three students read four books, where books is a bare
(nonspecific) noun, has only the collective reading (a total of four books); but with strong focus on the subject
numeral it can have either the collective or the distributive reading (a total of four books or a total of twelve), with
the distributive reading being the salient one (Kennelly 2004).
32. Sloppy identity involves some intricate problems (Hardt 2003). Focus on the (unellipsed) second pronoun forces
the sloppy reading (likes HIS students).
33. The precise mechanisms used to compute the desired meaning have been the subject of much debate (see
Chapter 5). Here is a similar example (Musan 1997) which depends on where in the subject phrase the focus falls:
‘SOME professors / Some PROFESSORS were happy in the sixties.’ With focus on the quantifier this means that
some of those who are professors today were happy in the sixties (when they were students). With focus on the
noun it means that some of those who were happy in the sixties were professors in the sixties (but not necessarily
today, since most of them are retired or deceased).
34. Brody (1990). Such a theory explains weak crossover, but encounters problems with constraints on extraction like
islands (Drubig 2003).
35. This type of lowering has actually been suggested for scrambling in Japanese (Bošković and Takahashi 1998).
36. Lee (2000), citing examples from Hindi. Similarly Russian allows OVS beside SVO except when inflectional
distinctions are neutralized (Jacobson 1998). In some languages, for instance Kanuri (Hutchinson 1986) and
colloquial Japanese (Kim 1995), case-marking affixes can be optionally dropped, which likewise is associated with
more rigid word order. Hindi also has a rule constraining contiguous nouns with identical case endings (Mohanan
1994), but that is probably a separate issue.
37. Szendrői (2003).
38. Weil (1869:89).
39. At least for genitive hyperbaton, a binary structure [inferiorem partem] [fluminis Rheni] is also required.
40. Kiss (1995).
41. Where it is helpful to distinguish finite and nonfinite forms of the verb, we will opt for assigning V to the finite
verb and a separate category (e.g. Ptcple) to the nonfinite
 INTRODUCTION 35

verb. An alternative would be to assign V to the nonfinite forms and I or T to the finite verb. We adopt a
traditional lexicalist approach to morphology.
42. Diesing (1992); Platzack (2000); Carlson (2003).
43. Rizzi (1997); Adger et al. (2004). It is quite common in Latin for the Topic in CP to be higher than the
complementizer. If the arguments are supposed to raise out of the VP, then the VP layer starts out as a complete
structured representation of the event with its thematic arguments, a projection of the lexical structure of the verb
into the syntax. Then one would get a more semantic characterization of the three layers, e.g. event (VP),
predication (IP), sentence type and discourse function (CP).
44. Rather than in its narrow sense of subject raising with verbs like seem, appear.
45. Morel (1927:44).
 1
ARGUMENTS OF VERBS

We will start by analyzing broad scope focus sentences, sentences answering the question ‘What happened?’ (clausal
scope) or at least ‘What did the subject do?’ (predicate or verb phrase scope). Such sentences stand the best chance of
revealing a neutral or default word order, from which more marked pragmatic structures will deviate in predictable
ways. Pending a more detailed analysis, we will refer to the neutral word order structure excluding the transitive subject
as the verb phrase. Since broad scope focus sentences are by definition less pragmatically articulated, neutral word
order should reflect general and intrinsic semantic and pragmatic properties of argument structure rather than
sentence specific informational structure. After a very brief section on subjects, we will concentrate on ditransitives,
oblique complements and adjuncts: the more noun phrases there are in a clause, the greater the number of possible
serializations.
At this stage in our argument we will simply deal in terms of serial order, leaving the discussion noticeably data heavy:
structural analysis and semantic interpretation will be put on hold until §1.5, which will be correspondingly theory
heavy and data light. For the time being, the terms “scrambling”1 and “topicalization” are used descriptively to refer to
the location of what is typically a weakly or, respectively, strongly topical constituent in a position to the left of its
default serial order position in the verb phrase.
There is little point in simply counting the frequency of the various serial orders. Raw statistics are not really
informative,2 since we do not know how much each of the two determinant factors (semantics and pragmatics) is
contributing to any particular serialization. It is well known that word orders can vary from one text to another, due
among other things to a different pragmatic mix. In fact even the most frequent word order can differ from one style
to another.3 Rather what we need to do is establish correlations between the various serializations and the pragmatic
structures they serve to encode. Note that pragmatic distinctions can be quite subtle and open to conflicting
interpretation, and that the information state at any point in the discourse can leave the speaker a choice of pragmatic
strategies. At the current stage of our knowledge about Latin word order, we decided it was better to stick with an
informal and
 ARGUMENTS OF VERBS 37

intuitive presentation rather than attempting either a statistically controlled scientific proof of our results or some form
of optimality theoretic analysis. Our support for the position that Latin has a neutral order is based on the empirical
evidence presented in this chapter. We are not assuming a priori either that grammatical relations are phrase
structurally encoded in all languages or that they are not structurally encoded in highly inflected free word order
languages.

1.1 SUBJECTS
In a typological survey of default word order in four hundred languages4 it emerged that the subject was initial
(preceding the object and the verb in either order) in 85% of the languages. So it is no surprise to find that in Latin too
the subject typically precedes the other arguments. Let's start with the proper name Caesar
26
Caesar eius dextram prendit (BG 1.20)
Caesar suas copias in proximum collem subducit (BG 1.22)
Caesar ad Lingonas litteras nuntiosque misit (BG 1.26)
Caesar Gallorum animos verbis confirmavit (BG 1.33)
Caesar singulis legionibus singulos legatos et quaestorem praefecit (BG 1.52)
Caesar postero die T. Labienum legatum… in Morinos… misit (BG 4.38)
Caesar in Belgis omnium legionum hiberna constituit (BG 4.38)
Caesar… ad flumen Tamesim in fines Cassivellauni exercitum duxit (BG 5.18)
Caesar exploratis regionibus albente caelo omnes copias castris educit (BC 1.68)
Caesar Germanos levis armaturae equitumque partem flumen traicit (BC 1.83).
There are many more examples. The subject is not necessarily in absolute initial position in the sentence, since
circumstantial expressions can easily be left adjoined to the sentence, like the temporal adverbials in the following
examples

26
Caesar took his right hand (BG 1.20). Caesar withdrew his forces to the nearest hill (BG 1.22). Caesar sent letters and messengers to the Lingones (BG 1.26). Caesar
strengthened the spirits of the Gauls by his words (BG 1.33). Caesar put the legates and the quaestor each in command of a legion (BG 1.52). The next day Caesar sent T.
Labienus, his legate, against the Morini (BG 4.38). Caesar set up the winter quarters of all the legions in the territory of the Belgae (BG 4.38). Caesar led his army into the
territory of Cassivelaunus, as far as the river Thames (BG 5.18). After scouting the area, at dawn Caesar leads his entire force out of camp (BC 1.68.) Caesar sends the light
armed Germans and part of his cavalry across the river (BC 1.83).
38 ARGUMENTS OF VERBS

27
Prima luce hostium equitatus ad castra accedit (BG 5.50)
Postridie eius diei Caesar… (BG 1.51; 2.12)
Sub vesperum Caesar… (BG 2.33)
Eodem fere tempore Caesar… (BG 3.28).
But apart from this sort of adjunction, there is still a substantial number of examples that do not have the inital subject.
These exceptions can involve scrambling or topicalization
28
munitiones institutas Caesar parat perficere (BC 1.83)
copias suas Caesar in proximum collem subducit (BG 1.24)
ex captivis Caesar cognovit Vercingetorigem.. castra movisse (BG 7.18)
In the first two examples (BC 1.83; BG 1.24) the direct object appears out of the neutral order and to the left of the
subject. In the third example (BG 7.18) the same applies to the prepositional phrase ex captivis. Final subjects can
appear in verb initial sentences (see Chapter 2)
29
Non respuit condicionem Caesar (BG 1.42)
Peragit concilium Caesar (BG 6.4)
Dimittit ad finitimas civitates nuntios Caesar (BG 6.34)
and in various other pragmatically marked structures which will be analyzed in later chapters. We have illustrated the
subject initial rule with a proper name, but it also applies to common noun phrases. Here are some examples with the
null head modifier5nostri ‘our men’
30
nostri acriter in hostes signo dato impetum fecerunt (BG 1.52)
nostri celeriter ad arma concurrunt (BG 5.39)
celeriter nostri clamore sublato pila in hostes immittunt (BG 6.8)
nostri omissis pilis gladiis rem gerunt (BG 7.88)
nostri… arma quae possunt adripiunt (BC 2.14)
nostri fortiter impetum eorum tulerunt (BC 3.37).
Once again scrambling and other marked orders also occur

27
At first light the enemy cavalry approached the camp (BG 5.50). On the day after Caesar… (BG 1.51). Towards evening Caesar… (BG 2.33). At about the same time
Caesar… (BG 3.28).
28
Caesar prepares to finish the fortifications he had begun (BC 1.83). Caesar withdraws his forces to the nearest hill (BG 1.24). From prisoners Caesar learned that Vercingetorix
had moved his camp (BG 7.18).
29
Caesar did not reject the proposal (BG 1.42). Caesar finished the conference (BG 6.4). Caesar sent messengers to the neighbouring states (BG 6.34).
30
When the signal was given our men fiercely attacked the enemy (BG 1.52). Our men quickly run to arms (BG 5.39). Quickly our men raise a cry and throw their javelins at the
enemy (BG 6.8). Having laid aside their javelins our men do their work with swords (BG 7.88). Our men snatch up what weapons they can (BC 2.14). Our men bravely
withstood their attack (BC 3.37).
 ARGUMENTS OF VERBS 39

31
impedimentis castrisque nostri potiti sunt (BG 1.26)
hanc si nostri transirent (BG 2.9)
ne navibus nostri circumvenirentur (BC 3.63).
As just noted, instead of appearing as a subject of predication at the left periphery of the clause, the grammatical
subject sometimes appears as a tail at the right periphery
32
ex grandibus saxis sex pedum murum… praeduxerant Galli (BG 7.46)
paulo longius progrediendum existimabat Caesar (BC 3.56)
Trepidantes… e media acie in extremam ad sinistrum cornu… agi iussit Hannibal (Livy 21.56.1)
perculsis acriter institerunt Romani (Livy 29.2.16)
Saltatorem appellat Murenam Cato (Pro Mur 13)
Pecunia mea tot annos utitur P. Quinctius (Pro Quinct 43)
Patrem occidit Sex. Roscius (Pro Rosc Am 39)
Non possum dicere planius quam ipse apud vos dixit Heius (Verr 2.4.27).
The last example (Verr 2.4) is a hyperbaton in which the subject fills two positions, a right peripheral tail position and a
left peripheral focus position.

Structural analysis
The above evidence is sufficient to show that the subject is mostly placed in a left peripheral position; subjects
preceded by left adjoined adverbials still count as initial; additionally, as expected in a free word order language,
subjects can appear in a variety of noninitial positions. In principle, this situation could arise in any or all of the
following ways. (i) The pragmatic theory: The subject is not associated with any particular position in the tree (such as
Spec IP), but is assigned to a number of different positions on the basis of its pragmatic function and/or referential
status;6 these positions are not reserved for subjects but can also be filled by other constituents having the required
pragmatic value. (ii) The grammatical theory (with movement): Whether the subject is overt or null depends directly on
the pragmatics, particularly topic continuity; many of the examples just cited involved subject switch. But when the
subject is overt, it is assigned to a grammatically defined subject position. This rule applies mutatis mutandis also in
subordinate clauses, in multiple argument nominalizations (§4.1), and in the multiple interrogative and indefinite
strings cited

31
Our men gained possession of the baggage and the camp (BG 1.26). If our men would cross this (BG 2.9). Lest our men be surrounded by the fleet (BC 3.63).
32
The Gauls had built a six foot wall out of boulders (BG 7.46). Caesar thought that he should advance a little farther (BC 3.56). Hannibal ordered the panicking (elephants) to
be driven from the center of the battle line to the outermost position on the left flank (Livy 21.56.1). The Romans vehemently attacked the overpowered (Spaniards) (Livy
29.2.16). Cato calls Murena a dancer (Pro Mur 13). Quinctius has been using my money for so many years (Pro Quinct 43). S. Roscius killed his father (Pro Rosc Am 39). I
cannot tell it more clearly than Heius himself told it before you (Verr 2.4.27).
40 ARGUMENTS OF VERBS

in (127) and (128) below. Although the subject is assigned to a single position, it can move to other positions to satisfy
(check) pragmatic requirements. (iii) The grammatical theory (without movement): The subject is assigned to a single
position and other constituents move around it. While all three theories contain elements of the truth, as just
formulated they are overly restrictive. The purely pragmatic approach of (i) probably goes too far in
degrammaticalizing the subject; and we probably need to allow for both subject movement as in (ii) and movement
of other constituents around the subject as in (iii).
When the subject is pragmatically a tail, it can show up in a right peripheral position, which may or may not be
dislocated.7 Under the assumptions of theory (ii), they could simply be stranded in the base verbal projection, or they
could be extraposed out of the nuclear clause. Under the assumptions of theory (iii), the verb phrase could raise in its
entirety to the left of the subject in its regular initial position. Note that, on this latter approach, the pragmatic value of
the subject is not cued by which position the subject docks in but by whether a nonsubject constituent moves across
the subject. This is less direct and loses the generalization that different positions in the syntax host different pragmatic
categories.
Some evidence will be cited in later chapters (e.g. §3.1) pointing to two different left edge subject positions, a higher
subject and a lower subject (one position in the serial order can translate into more than one position in the structural
analysis). This is one way of accounting for celeriter nostri arma cepissent (BG 5.26) versus Nostri celeriter arma ceperunt cited
in (143) and (144) below. Medial subjects could arise by raising another constituent across the subject either in the
lower subject position only or in both the higher and the lower positions.

1.2 DITRANSITIVES
Ditransitive verbs take two complements, a direct and an indirect object. They include property transfer verbs and
various prepositional prefix verbs. On this narrow definition of the class of ditransitives, they are distinguished from
other trivalent (triadic, three-argument) verbs which take a direct object and an oblique (see §1.3), as well as from those
that take two direct objects (like docere ‘teach’).

Dare
For our first data set we will look at the serial order of direct and indirect object with the basic verb of giving dare ‘give.’

Neutral Order
In the neutral order, the direct object (DO) precedes the indirect object (IO). Here are some examples from Cicero8
 ARGUMENTS OF VERBS 41

33
de eorum sententia leges Halaesinis dedit (Verr 2.2.122)
pecuniam Staieno dedit Oppianicus (Pro Clu 84)
agros locupletium plebi… colendos dedit (De Rep 3.16)
Philotes Alabandensis hypothecas Cluvio dedit (Ad Fam 13.56.2)
Hic di immortales… mentem illi perdito ac furioso dederunt (Pro Mil 88)
est pollicitus se venenum regi daturum (De Off 1.40.).
The same order is amply attested in Livy
34
celebre ad posteros nomen flumini dedit (Livy 1.3.9)
praedam omnem suo tantum militi dedit (Livy 3.29.1)
rebellandi causam Samnitibus dedit (Livy 9.21.3)
signum ad invadendos hostes equitibus dedit (Livy 10.29.11)
uterque… consul captum oppidum diripiendum militi dedit (Livy 10.44.2)
transitum hostibus dedit (Livy 26.6.2)
Postremo praefecturam eius filio suo dedit (Livy 26.40.6)
victoriam haud dubiam Syphaci dedit (Livy 29.33.5)
leges Macedoniae dedit (Livy 45.32.7.)
quia is victor pacem Aequis dederat (Livy 3.2.3)
qui obsides Scipioni dederat (Livy 21.61.5)
quia obsides Larisaeis dederant (Livy 42.53.7)
locum adversae factioni dederant ad Popilium… accersendum (Livy 43.22.3)
ii regem Celtico dabant (Livy 5.34.2)
tutum… receptum sociis… dabant (Livy 42.59.5).
This DO – IO order is grammatically determined in broad scope focus. It does not depend on definiteness or on the
informational status of the argument phrases. There is no simple and direct correlation between this neutral order and
any particular sequence of pragmatic categories

33
According to their judgment gave laws to the Halaesini (Verr 2.2.122). Oppianicus gave the money to Staienus (Pro Clu 84). Gave the cultivation of the lands of the rich to the
plebs (De Rep 3.16). Philotes of Alabanda has given Cluvius a mortgage (Ad Fam 13.56.2). At this point the immortal gods gave that depraved and insane man the idea (Pro
Mil 88). Promised to give poison to the king (De Off 1.40).
34
Gave the river the name current in posterity (Livy 1.3.9). Gave all the spoils just to his own soldiers (Livy 3.29.1). Gave the Samnites the pretext for starting the war again
(Livy 9.21.3). Gave the cavalry the signal to charge the enemy (Livy 10.29.11). Each consul gave the city he had captured to the soldiers to be plundered (Livy 10.44.2). Gave
the enemy passage (Livy 26.6.2). Finally gave his own son that man's command (Livy 26.40.6). Gave Syphax a by no means doubtful victory (Livy 29.33.5). Gave laws to
Macedonia (Livy 45.32.7). Because this victor had given the Aequi peace (Livy 3.2.3). Who had given Scipio hostages (Livy 21.61.5). Because they had given the Larisaeans
hostages (Livy 42.53.7). Had given the opposing faction the opportunity to summon Popilius (Livy 43.22.3). They used to give the Celtic nation its king (Livy 5.34.2). Were
giving the allies safe retreat (Livy 42.59.5).
42 ARGUMENTS OF VERBS

35
senatus libertatem his civitatibus dedit (Livy 33.34.10)
praedam custodiendam… trecentis Cretensium dedit (Livy 42.65.5)
triplexque stipendium equitibus dederat (Livy 5.12.12)
Haec spem ad resistendum oppidanis dabant (Livy 43.19.9)
quo tempore P. Valerius… arma plebi dedit (Livy 3.20.3)
praedam militi dedit (Livy 6.2.12).
In the first example (Livy 33.34) the direct object is new information and the indirect object is old information;
whereas in the second example (Livy 42.65) the direct object is old information and the indirect object is new
information. In the third example (Livy 5.12) both direct and indirect objects are new information. In the fourth and
fifth examples (Livy 43.19; 3.20) the indirect object is tail information; in the last example (Livy 6.2) it is a focus.

Indirect object movement

Deviation from the neutral order is due to topicalization or scrambling of the indirect object to the left of the direct
object. Movement is overtly demonstrable when the landing site of the indirect object is above the subject
36
tantum modo aratoribus Metellus obsides non dedit (Verr 2.3.124)
Siciliae civitatibus bello fugitivorum M'. Aquilius etiam mutuum frumentum dedit (De Leg Agr 2.83)
C. Mario L. Valerio consulibus senatus rem publicam defendendam dedit (Phil 8.15)
legatis quaestores sumptum quem opportebat dari non dederunt (De Inv 2.87).
Scrambling may also be assumed (without being unequivocally demonstrable) where the neutral order is inverted
37
Faliscis pacem petentibus annuas indutias dedit (Livy 10.46.12)
pavor Numidarum Romanis iam admodum fessis victoriam dedit (Livy 21.29.3)
prior fessis stando signum receptui dabat (Livy 28.14.3)

35
The Senate gave freedom to these states (Livy 33.34.10). Gave the spoils to three hundred Cretans to guard (42.65.5). Had given triple pay to the cavalrymen (Livy 5.12.12).
These things gave the townspeople hope of resisting (Livy 43.19.9). When P. Valerius gave arms to the plebs (Livy 3.20.3). Gave the spoils to the soldiers (Livy 6.2.12).
36
Metellus has just about given hostages to the farmers (Verr 2.3.124). During the Servile War Manius Aquilius even gave grain on loan to the cities of Sicily (De Leg Agr 2.83),
The Senate gave the defence of the republic to the consuls C. Marius and L. Valerius (Phil 8.15). The financial authorities did not give the ambassadors the expense allowance
which ought to have been given (De Inv 2.87).
37
He gave a year's time to the Faliscans seeking peace (Livy 10.46.12). The panic of the Numidians gave victory to the now very exhausted Romans (Livy 21.29.3). He was the
first to give the signal for withdrawal to those exhausted from standing (Livy 28.14.3).
 ARGUMENTS OF VERBS 43

38
pictori quam vellet eligendi potestatem dederunt (De Inv 2.3)
qui Graeciae formam rerum publicarum dederunt (Tusc 2.36)
Lacedaemonii regibus suis augurem adsessorem dederunt (De Div 1.95).
However, in many of these examples the direct object is an abstract noun, which might more readily semantically
incorporate into the verb, pointing to an additional preverbal position for nonspecific indefinites. This is also the case
in the set phrase negotium dare ut which commonly occurs with preverbal direct object
39
viginti tribunis militum negotium dederunt ut… (Livy 3.51.10)
Senatus Cn. Servilio consuli negotium dedit ut… (Livy 44.18.5)
M. Claudio clienti negotium dedit ut… (Livy 3.44.5)
senatus C. Scribonio negotium dedit ut… (Livy 35.6.5).
Contrastive topics appear to the left of the direct object
40
ea quae gignantur e terra… bestiis autem sensum et motum dedit (De Nat Deor 2.34)
Termesso pacem dedit,… item Aspendiis (Livy 38.15.6)
Clazomeniis… et Drymussam insulam dono dederunt, et Milesiis… (Livy 38.39.9).
This also applies in instances of chiasmus
41
militibus de praeda quinquagenos denarios dedit, duplex centurionibus (Livy 40.43.7)
quae quidem nostris amicis… aditum ad caelum dederunt, libertatem populo Romano non dederunt (Ad Att
14.14.3).

Direct object movement

Like indirect objects, direct objects too can be topicalized to the left of a subject. This does not alter the relative serial
order of direct and indirect object, but obviously it does alter the structural position of the former

38
Gave the painter the opportunity of selecting the one that he wished (De Inv 2.3). Who gave Greece the form of its states (Tusc 2.36). The Lacedaemonians gave an augur to
their kings as a counselor (De Div 1.95).
39
They gave the twenty military tribunes the task of… (Livy 3.51.10). The Senate gave the consul Cn. Servilius the task of… (Livy 44.18.5). He gave M. Claudius, his client, the
task of… (Livy 3.44.5). The Senate gave C. Scribonius the task of… (Livy 35.6.5).
40
Those things which are produced from the earth… to animals, however, gave sensation and motion (De Nat Deor 2.34). Gave peace to Termessus… the same to the
Aspendians (Livy 38.15.6). To the Clazomenians… gave also the island of Drymussa as a gift, and to the Milesians… (Livy 38.39.9).
41
To the infantry soldiers he gave fifty denarii from the spoils, double to the centurions (Livy 40.43.7). Which to our friends at least gave admittance to heaven, but not freedom
to the Roman people (Ad Att 14.14.3).
44 ARGUMENTS OF VERBS

42
urbanum veterem exercitum Fulvius consul C. Fulvio Flacco legato in Etruriam dedit ducendum (Livy 27.8.12);
The position of the adverbials (paene, eodem die) in the following examples suggests that the direct object can also be
scrambled out of the verb phrase
43
pons sublioius iter paene hostibus dedit (Livy 2.10.2)
Hostilius et Furius damnati praedes eodem die quaestoribus urbanis dederunt (Livy 38.58.2).
Direct object phrases with a demonstrative have a topical flavour which would fit well with string vacuous scrambling
(scrambling that does not change the overt word order)
44
iam ego hanc mactatam victimam, si modo sancti quicquam in terris esse di volunt, legatorum manibus dabo (Livy
4.19.4)
iam ego hanc victimam manibus peremptorum foede civium dabo (Livy 22.6.4).
In quite a few instances the direct object is a contrastive topic, indicating string vacuous movement
45
bovem eximium Marti immolavit, centum boves militibus dono dedit (Livy 7.37.3)
qui regna quasi praedia tetrarchis, qui immanis pecunias paucis dederunt (Ad Att 2.9.1)
parte exercitus consul castra Aequorum oppugnabat; partem Tusculanis dederat (Livy 3.23.4).
In the following example all three arguments are contrastive
46
plebes consulatum L. Sextio… dedit; patres praeturam Sp. Furio… gratia campestri ceperunt (Livy 7.1.2).
Both direct and indirect object can appear to the left of the subject, indicating that both have moved from their neutral
positions

42
The consul Fulvius gave the old city army to C. Fulvius Flaccus, his legate, to lead into Etruria (Livy 27.8.12).
43
The Sublician Bridge almost gave passage to the enemy (Livy 2.10.2). Hostilius and Furius, the convicted, gave bond the same day to the city quaestors (Livy 38.58.2).
44
Now I will give this sacrificial victim to the spirits of the ambassadors, if the gods wish there to be anything sacred on earth (Livy 4.19.4). Now I will give this victim to the
spirits of the citizens foully killed (Livy 22.6.4).
45
The choice ox he sacrificed to Mars, the hundred oxen he gave as a gift to the soldiers (Livy 7.37.3). Who have given kingdoms to tetrarchs like private estates, and immense
amounts of money to a few (Ad Att 2.9.1). With part of the army the consul besieged the camp of the Aequi; a part he had given to the Tusculans (Livy 3.23.4).
46
The plebeians gave a consulship to L. Sextius…; the patricians obtained a praetorship for Spurius Furius by their influence in the Campus Martius (Livy 7.1.2).
 ARGUMENTS OF VERBS 45

47
recessumque primis ultimi non dabant (BG 5.43)
imperium C. Caesari belli necessitas, fascis senatus dedit (Phil 11.20)
si enim rationem hominibus di dederunt, malitiam dederunt (De Nat Deor 3.75)
tum Segoveso sortibus dati Hercynei saltus; Belloveso haud paulo laetiorem in Italiam viam di dabant (Livy 5.34.4).
There are a numer of examples of hyperbaton in which components of the direct object phrase occupy two different
positions at the same time. The structure becomes syntactically overt because the indirect object is scrambled and the
direct object wraps around the indirect object
48
aditum petentibus conveniundi non dabat (Nepos 4.3.3)
locum hostibus introeundi dedit (Jug 38.6)
potestatem Pompeio civitatem donandi dederat (Pro Balb 32)
propinquitas castrorum celerem superatis ex fuga receptum dabat (BC 1.82)
Hanc vos igitur… tribuno plebis potestatem dabitis ut… (De Dom 44)
coronam auream consuli centum et quinquaginta pondo dederunt (Livy 38.9.13)
precibus eventum vestris senatus quem videbitur dabit (Livy 6.26.2).
In the first three examples (Nepos 4.3; Jug 38.6; Pro Balb 32), the head of the object phrase is in a scrambled position
(as also the indirect object) and its complement is a focus. In the next two examples (BC 1.82; De Dom 44) the
modifier is in a focus position and the head of the object phrase is stranded in the neutral position. In the last two
examples (Livy 38.9; 6.26) the head of the object phrase is scrambled to the left of the subject and its modifier is a
focus.

Other verbs of giving

Verbs of giving with a richer lexical content like tradere ‘hand over,’ reddere ‘give back,’ tribuere ‘devote, attribute’ confirm
the evidence of the dare data set. The neutral order DO–IO predominates with all three verbs. We will start with
examples for tradere

47
The rear would not give retreat to the van (BG 5.43). The necessity of war gave C. Caesar his command, the Senate his fasces (Phil. 11.20). For if the gods have given men
reason, they have given malice (De Nat Deor 3.75). Then the Hercynean mountain woodlands were given by lot to Segovesus; the gods gave to Bellovesus a much happier
road, to Italy (Livy 5.34.4).
48
Would not give the opportunity of meeting to those petitioning (Nepos 4.3.3). Gave a place of entry to the enemy (Jug 38.6). Had given Pompey the power of granting
citizenship (Pro Balb 32). The proximity of the camps gave the defeated a quick refuge from flight (BC 1.82). Will you therefore give a tribune of the plebs the power to… (De
Dom 44). Gave the consul a golden crown weighing one hundred and fifty pounds (Livy 38.9.13). The senate will give your entreaties the answer which seems right to it (Livy
6.26.2).
46 ARGUMENTS OF VERBS

49
naves… Cleomeni tradit (Verr 2.5.82)
quadraginta navium classem Himilconi tradit (Livy 22.19.3)
stipendium quaestoribus, frumentum aedilibus, captivos Fulvio praetori tradit (Livy 23.41.7)
bona patria fortunasque eius Bidinis tradidit (Verr 2.2.59)
aratorem decumano tradidit (Verr 2.3.20)
si tu totam rempublicam nefariis latronibus tradidisses (In Pis 57)
qui exercitum hostibus populi Romani tradidit (De Orat 2.164)
cum Sardiniam legionemque Ti. Claudio tradidisset (Livy 29.13.5)
vix dimidium militum quam quod acceperat successori tradiderit (Livy 35.1.2)
insulam Achaeis tradidit (Livy 36.32.2)
Baebius exercitum M. Pinario… tradiderat (Livy 40.25.8).
Here are the neutral order examples for reddere
50
captivorum circiter xx milia Haeduis Arvernisque reddit (BG 7.90)
M. Gallius Q. f. mancipia Sallustio reddidit (Ad Att 11.20.2)
urbem agros suaque omnia… Reginis reddidimus? (Livy 31.31.7)
etsi corpus patri reddiderit (Livy 39.47.10),
and finally those for tribuere
51
veniam tamen aliquam dolori meo tribueretis (De Prov 1)
ampliorem honorem alteri tribuebat (De Prov 27)
praefectos subsidiis attribuerat (Livy 10.40.7)
Lacedaemoniorum victorias culpae suae tribuebant (Nepos 7.6.2)
omnia reliqua tempora aut litteris aut Atheniensium rei publicae tribueret (Nepos 25.4.3).
There are examples of the indirect object moving to the left of the direct object for all three verbs

49
Hands over the ships to Cleomenes (Verr 2.5.82). Handed over the fleet of forty ships to Himilco (Livy 22.19.3). Handed over the tribute to the quaestors, the grain to the
aediles, and the prisoners to Fulvius, the praetor (Livy 23.41.7). Handed over to the Bidini his inheritance and fortune (Verr 2.2.59). Handed the farmer over to the tax
collector (Verr 2.3.20). If you handed over the whole state to wicked plunderers (In Pis 57). Who handed over the army to the enemies of the Roman people (De Orat 2.164).
When he had handed over Sardinia and the legion to Ti. Claudius (Livy 29.13.5). Handed over to his successor hardly half the soldiers he had received (Livy 35.1.2). Handed
over the island to the Achaeans (Livy 36.32.2). Baebius had handed over the army… to M. Pinarius (Livy 40.25.8).
50
Gives back about twenty thousand prisoners to the Haedui and Arverni (BG 7.90). M. Gallius, son of Quintus, has given back slaves to Sallustius (Ad Att 11.20.2). Did we
give back to the people of Rhegium their city, their lands, and all their goods? (Livy 31.31.7). Even if it had given his body back to his father (Livy 39.47.10).
51
Nevertheless you would grant some pardon for my resentment (De Prov 1). He would grant greater honour to another (De Prov 27). Had assigned commanders to the
reserves (Livy 10.40.7). They attributed the Lacedaemonians' victories to their own faults (Nepos 7.6.2). Devote all the rest of his time either to letters or to the Athenian state
(Nepos 25.4.3).
 ARGUMENTS OF VERBS 47

52
isdemque custodiam navium longarum tradidit (BC 3.39)
Aegyptum profugisse atque Aegyptiis leges et litteras tradidisse (De Nat Deor 3.56)
cum praetori exercitum tradidisset (Livy 22.57.1)
quae Hannibali Locros tradiderat (Livy 29.6.5)
senatusque et populus Romanus Thermitanis… urbem agros legesque suas reddidisset (Verr 2.2.90)
perdiserte populo rationem operis sui reddidisse (De Orat 1.62)
qui amori auctoritatem tribueremus (Tusc 4.71)
qui mortuis tam religiosa iura tribuerunt (De Amic 13).
The general pattern in these examples is for the focus to be on the direct object, while the scrambled indirect object is
relatively topical. This is the typical pattern in topicalization and scrambling. Here is an example in which the scrambled
indirect object scopes over conjoined object–verb structures
53
clarissimis ducibus supplicationum honorem tribuemus, imperatorium nomen adimemus? (Phil 14.12).
For the sake of convenience we shall refer to such structures by the potentially misleading term “left node raising.”9
When the indirect object scopes over the conjuncts, it is raising to the left node, when the direct object scopes over the
conjuncts it is raising of the left node (assuming that the interpretation is not via a null anaphoric pronoun or ellipsis).
As already noted, direct object scrambling does not change the serial order, but it can be discerned when some other
word or constituent intervenes between the direct and indirect object and probably also when the direct object has a
linking anaphoric demonstrative
54
exercitumque ad campos Macros consuli tradidit (Livy 41.18.6)
cultrum deinde Collatino tradit, inde Lucretio ac Valerio (Livy 1.59.2)
hiberna cum legato praefectoque tuo tradidisses (In Pis 86)
liberos quoque parvos regios Ion… Octavio tradidit (Livy 45.6.9)

52
He handed over to them the custody of the warships (BC 3.39). To have fled to Egypt and to have given laws and letters to the Egyptians (De Nat Deor 3.56). When he had
handed over the army to the praetor (Livy 22.57.1). Which had handed over Locri to Hannibal (Livy 29.6.5). And the Senate and the Roman people had given back to the
Thermitani their city, their lands, and their laws (Verr 2.2.90). Most eloquently gave an account of his work to the people (De Orat 1.62). In order to attribute authority to love
(Tusc 4.71). Who devoted such pious rites to the dead (De Amic 13).
53
Shall we award to the most illustrious generals the honour of thanksgiving ceremonies, but take away from them the name of commander (Phil 14.12).
54
He handed over the army to the consul at Campi Macri (Livy 41.18.6). Then he hands the knife over to Collatinus, and then to Lucretius and Valerius (Livy 1.59.2). When you
had handed over your winter quarters to your legate and prefect (In Pis 86). Ion also handed over the small royal children to Octavian (Livy 45.6.9).
48 ARGUMENTS OF VERBS

55
hunc honorem mulieri Larentiae tribuerunt (Ad Brut 23.8)
Eos nummos tamen iste Archagatho non reddidit (Verr 2.4.53).
Finally, here is an instance of the double contrast type with neutral order in both components
56
laudem veritati tribuebas, crimen gratiae concedebas (Pro Rosc Com 19);
similarly with other verbs
57
spem improbis ostendistis, timorem bonis iniecistis (De Leg Agr 1.23).

Donare
Like English to present, Latin donare is used in two constructions. In one construction the transferred property is in the
accusative and the recipient is in the dative; in the other the recipient (presentee) is in the accusative and the transferred
property (thing presented) is in the instrumental ablative. According to the terminology we have adopted, both
constructions are trivalent but only the former is ditransitive; most of our examples belong to the latter type, but for
ease of comparison we will analyze both types here. Irrespective of which construction is chosen, the neutral order is
accusative first
58
mercedes habitationum annuas conductoribus donavit (BC 3.21)
agellos… militi suo donavit (Sen De Ben 5.24.3)
Noster hic Magnus… nonne Theophanem Mytilenaeum… in contione militum civitate donavit…? (Pro Arch 24)
Is igitur Iguvinatem M. Annium Appium… civitate donavit (Pro Balb 46)
P. Decium patrem tribunum militum frondea donavit (Pliny NH 16.11)
scribas suos anulis aureis in contione donarunt (Verr 2.3.185)
ut libertum suum Asiaticum equestri dignitate donaret (Tac Hist 2.57).
In the following examples the presence of intervening material indicates object scrambling or topicalization

55
Bestowed this honour on the lady Larentia (Ad Brut 23.8). Nevertheless he did not give this money back to Archagathus (Verr 2.4.53).
56
You bestowed praise on truth, but you accusation was a concession to obligation (Pro Rosc Com 19).
57
You have offered hope to the wicked, instilled fear in the good (De Leg Agr 1.23).
58
Presented the annual rent for lodgings to the renters (BC 3.21). Presented small plots to his soldiers (Sen De Ben 5.24.3). Did not our own Pompey in an assembly of his
soldiers present Theophanes of Mytilene with citizenship? (Pro Arch 24). He therefore presented M. Annius of Iguvium with citizenship (Pro Balb 46). Presented P. Decius
père, the military tribune, with a garland of leaves (Pliny NH 16.11). They presented their scribes with gold rings in the assembly (Verr 2.3.185). To present his freedman
Asiaticus with equestrian rank (Tac Hist 2.57).
 ARGUMENTS OF VERBS 49

59
cohortemque postea… cibariis militaribusque donis amplissime donavit (BC 3.53)
Cn. Pompeius pater… P. Caesium… nonne civitate donavit? (Pro Balb 50)
Ob id Aelium Thurini statua et corona aurea donarunt. (Pliny NH 34.32).
It is also possible for the oblique to be scrambled to the left of the direct object
60
qui civitate multos donavit (Pro Arch 26)
Nero civitate Romana ambos donavit (Tac Ann 13.54).
The focus is on the object quantifiers in both examples.

Adimere
Turning now to verbs of depriving, we will start with adimere ‘take away from,’ which usually takes accusative of the
property transferred and dative (indirect object) of the person from whom the property is taken. Here are some
examples of the neutral order
61
primum ut aliquid Caesari adimat, inde ut aliquid Pompeio tribuat (Ad Fam 8.10.3)
consul… arma omnibus cis Hiberum Hispanis adimit (Livy 34.17.5)
si agrum Campanis ademissent (De Leg Agr 2.88)
agrumque civibus ademissent (De Leg Agr 2.90)
Vectigalia Iuliana Lupercis ademistis (Phil 13.31)
signa alicui manipulo aut cohorti ademisset? (Livy 27.13.7)
arma Satricanis ademit (Livy 9.16.10)
nisi di mentem regi ademissent (Livy 44.6.14).
The indirect object can be scrambled to the left of the direct object
62
Lutarius Macedonibus… tris lembos adimit (Livy 38.16.6)
iste infanti pupillae fortunas patrias ademit (Verr 2.1.153)
Populus Romanus… municipiis civitatem ademit (De Dom 79)

59
And afterwards he made the cohort a most generous presentation of provisions and military gifts (BC 3.53). Did not Cn. Pompeius père present P. Caesius with citizenship?
(Pro Balb 50). On account of this the Thurini presented Aelius with a statue and gold crown (Pliny NH 34.32).
60
Who endowed many with citizenship (Pro Arch 26). Nero presented both with Roman citizenship (Tac Ann 13.54).
61
First to take something away from Caesar, then to bestow something on Pompey (Ad Fam 8.10.3). The consul takes arms away from all the Spaniards on this side of the Ebro
(Livy 34.17.5). If they took their land away from the Capuans (De Leg Agr 2.88). Took away land from the citizens (De Leg Agr 2.90). You have taken away the Julian revenues
from the Luperci (Phil 13.31). Would have taken away the standards from any maniple or cohort? (Livy 27.13.7). Took arms away from the Satricani (Livy 9.16.10). If the gods
had not taken away reason from the king (Livy 44.6.14).
62
Lutarius takes away three fast boats from the Macedonians (Livy 38.16.6). That man took away from the infant ward her inheritance from her father (Verr 2.1.153). The
Roman people took away citizenship from municipalities (De Dom 79).
50 ARGUMENTS OF VERBS

63
cum M. Antistio Pyrgensi equum ademisset (De Orat 2.287)
di immortales et vestris et hostium imperatoribus mentem ademerunt (Livy 9.9.19)
cum signifero signum ademisset (Livy 25.14.7)
multis equos ademerunt (Livy 43.16.1).
Both orders are found in what we have loosely been calling left node raising structures
64
clarissimis ducibus supplicationum honorem tribuemus, imperatorium nomen adimemus? (Phil 14.12)
imperium navium legato populi Romani ademisti, Syracusano tradidisti (Verr 2.5.137)
Idem Cretensibus… spem deditionis non ademit obsidesque imperavit (Pro Leg Man 35)
tribunis plebis sua lege iniuriae faciendae potestatem ademerit, auxilii ferendi reliquerit (De Leg 3.22).
Left node raising is not compulsory
65
edictoque suo non luctum patribus conscriptis sed indicia luctus ademerint (Pro Planc 87)
Illi aditum litoris Syracusanis ademerunt, tu imperium maritimum concessisti (Verr 2.5.85).
These examples just have the neutral order. One or both arguments can also be scrambled to a position preceding the
subject
66
consulari homini P. Clodius… civitatem adimere potuit (De Dom 79)
iumenta… Gallo abigenti duo milites Romani ademerunt (Livy 7.14.4).

63
When he had taken a horse away from M. Antistius of Pyrgi (De Orat 2.287). The immortal gods took away reason from both your commanders and those of the enemy (Livy
9.9.10). When he had taken the standard away from the standard-bearer (Livy 25.14.7). They took away their horses from many of them (Livy 43.16.1).
64
Shall we award the honour of thanksgiving ceremonies to the most illustrious generals, but take away from them the name of commander? (Phil 14.12). You have taken away
command of the ships from the legate of the Roman people and handed it over to a Syracusan (Verr 2.5.137). Likewise he did not take away hope of surrender from the
Cretans, but demanded hostages of them (Pro Leg Man 35). By his law took away the power of causing injury from the tribunes of the plebs and left them only that of bringing
assistance (De Leg 3.22).
65
By their edict they took away not mourning from the senators but the tokens of mourning (Pro Planc 87). They took away access to the coast from the Syracusans, you
conceded command of the sea (Verr 2.5.85).
66
Was P. Clodius able to take away citizenship from a man of consular rank? (De Dom 79). Two Roman soldiers took the beasts of burden away from a Gaul who was driving
them off (Livy 7.14.14).
 ARGUMENTS OF VERBS 51

Extorquere and eripere

For verbs of depriving with a richer lexical content we chose extorquere ‘wrench away from’ and eripere ‘snatch from,’
which take an accusative of the transferred property, a dative of the person from whom the property is taken and/or
an ablative of the source from which it is taken (with or without a preposition). The neutral order is direct object –
indirect object – source. In the following examples the direct object precedes the other (lexical) argument(s)
67
quod ei ferrum e manibus extorsimus (Cat 2.2)
qui numquam sententias de manibus iudicum vi quadam orationis extorsimus (De Orat 2.74)
victoriam hosti extorqueamus, confessionem erroris civibus (Livy 22.29.2)
Postumium Pyrgensem suffragium populo Romano extorsisse (Livy 25.4.4.)
C. Marium e civili ferro atque ex impiis manibus eripuerunt (Pro Planc 26)
paene victam rem publicam ex manibus hostium eripuit (Ad Fam 12.13.1)
ut inde ex propinquo… Capuam Romanis eriperent (Livy 9.27.3)
ut urbem Romanam e manibus hostium eriperent (Livy 26.9.8)
coloniam ex hostibus eripere (Livy 41.14.4).
Scrambling of an argument to the left of the direct object is well documented
68
mancipes a civitatibus pecunias extorserunt (Verr 2.3.175)
Nos qui P. Lentulo ferrum et flammam de manibus extorsimus (Pro Flacc 97)
patribus omnem honorem eripuit (De Leg 3.19)
legato tuo viaticum eripuerunt (Ad Fam 12.3.2).
An argument can also move to a position preceding the subject

67
Because we wrenched his sword out of his hands (Cat 2.2). Who never wrenched away judgments from the hands of jurors by some force of rhetoric (De Orat 2.74). Let us
wrench away victory from the enemy and confession of error from the citizens (Livy 22.29.2). That Postumius of Pyrgi had wrenched away the right to vote from the Roman
people (Livy 25.4.4). Snatched C. Marius from civil war and impious hands (Pro Planc 26). Snatched the nearly conquered republic from the hands of the enemy (Ad Fam
12.13.1). So that from near there they might snatch Capua from the Romans (Livy 9.27.3). So that they might snatch the Roman city from the hands of the enemy (Livy 26.9.8).
To snatch the colony from the enemy (Livy 41.14.4).
68
The agents snatched away the money from the cities (Verr 2.3.175). We who wrenched away sword and fire out of P. Lentulus' hands (Pro Flacc 97). Snatched every honour
from the senators (De Leg 3.19). They have snatched the travel allowance from your legate (Ad Fam 12.3.2).
52 ARGUMENTS OF VERBS

69
Volaterranis… L. Sulla… civitatem eripere non potuit (De Dom 79)
ni patribus tribuni cum iure ac maiestate dempta animos etiam eripuerint (Livy 4.2.14)
Pater… naturae concessit. Fratri… propincus per scelus vitam eripuit (Jug 14.15)
ut Veios… omnes Etruriae populi ex obsidione eriperent (Livy 5.17.7).
Various manifestations of the left node raising structure are illustrated in the following examples
70
tu populo Romano subsidia belli, tu ornamenta pacis eripias? (De Leg Agr 1.3)
Porcia lex libertatem civium lictori eripuit, Labienus… carnifici tradidit (Pro Rab Perd 12)
ut huius quoque generis laudem iam languenti Graeciae eripiant et transferant in hanc urbem (Tusc 2.5).

Prepositional prex verbs

Praecere
As our first example of this category of ditransitive we chose praeficere ‘put in charge of.’ The neutral order for praeficere
is again direct object before indirect object
71
Itaque consules… T. Maenium dilectui habendo praefecerunt (Livy 39.20.4)
cur fratrem provinciae non praefecissem (Ad Att 7.1.1)
tum seditiosos homines rei publicae praeficiebant (Pro Flacc 16)
Amynander Philippum Megalopolitanum insulae praefecit (Livy 36.31.12)
Crassum Samarobrivae praeficit (BG 5.47)
T. Quinctium Crispinum in eius locum classi castrisque praeficit veteribus (Livy 24.39.13).

69
L. Sulla was not able to snatch citizenship from the Volaterrans (De Dom 79). Unless the tribunes had snatched from the senators even their spirits along with the rights and
powers they had taken away (Livy 4.2.14). My father has died. From my brother a relative has criminally snatched his life (Jug 14.15). So that all the peoples of Etruria might
rescue Veii from siege (Livy 5.17.7).
70
Are you to snatch from the Roman People their support in war, their ornaments in peace (De Leg Agr 1.3). The Porcian Law took control over the freedom of citizens away
from the lictor, Labienus has handed it over to the executioner (Pro Rab Perd 12). That they may snatch glory of this kind too from a decadent Greece and transfer it to this
city (Tusc 2.5).
71
Therefore the consuls put T. Maenius in charge of recruitment (Livy 39.20.4). Why I had not put my brother in charge of the province (Ad Att 7.1.1). Then they would put
seditious men in charge of the republic (Pro Flacc 16). Amynander put Philip of Megalopolis in charge of the island (Livy 36.31.12). He puts Crassus in charge of Samarobriva
(BG 5.47). In his place he puts T. Quinctius Crispinus in charge of the fleet and the old camp (Livy 24.39.13).
 ARGUMENTS OF VERBS 53

The neutral order is used for a simple report of the situation; note cur in the second example (Ad Att 7.1) and tum in
the third (Pro Flacc 16). There are also some cases of this order where the direct object is a contrastive topic and so
presumably has been string vacuously moved to a higher position
72
Cassium sibi legavit, Brutum Galliae praefecit, Sulpicium Graeciae (Ad Fam 6.6.10)
Crassum… Sabinum… D. Brutum adulescentem classi Gallicisque navibus… praeficit (BG 3.11)
eundem Achillam cuius supra meminimus omnibus copiis praefecit (BC 3.108)
legatum alterum P. Sulpicium equitibus praeficiunt (Livy 3.70.2)
Geminum Servilium… legioni Romanae… praeficiunt (Livy 22.40.6).
Topicalization of the indirect object is common and easily detectable from the inverted serial order
73
et huic procurationi certum magistratum praefecerat (De Leg 2.66)
ei munitioni quam fecerat T. Labienum legatum praefecit (BG 1.10)
ei praesidio navibusque Q. Atrium praefecit (BG 5.9)
ei legioni castrisque Q. Tullium Ciceronem praefecit (BG 6.32)
his castris Curionem praefecit (BC 1.18)
eique negotio Q. Fufium Calenum legatum praefecit (BC 1.87)
eique rei M. Scaurum praefecit (De Har Resp 43).
In these examples the recurrent demonstrative shows that movement is triggered by the topical status of the indirect
object. Topicalization is not obligatory in the presence of a demonstrative
74
Brutum adulescentem his copiis praeficit (BG 7.9).
Indirect object movement also occurs without a demonstrative
75
dextro cornu L. Volumnium, sinistro L. Scipionem, equitibus legatos alios… praefecit (Livy 10.40.7)

72
Cassius he made his legate, Brutus he put in charge of Gaul, Sulpicius of Greece (Ad Fam 6.6.10). Crassus… Sabinus… D. Brutus the younger he puts in charge of the fleet
and the Gallic ships (BG 3.11). He put the same Achilles, whom we mentioned above, in charge of all the forces (BC 3.108). The other legate, P. Sulpicius, they put in charge of
the cavalry (Livy 3.70.2). They put Geminus Servilius in charge of the Roman legion (Livy 22.40.6).
73
He put a specific magistrate in charge of this enforcement (De Leg 2.66). He put T. Labienus in charge of this fortification which he had built (BG 1.10). He put Q. Atrius in
charge of the garrison and the ships (BG 5.9). He put Q. Tullius Cicero in charge of the legion and the camp (BG 6.32). He put Curio in charge of this camp (BC 1.18). He put
Q. Fufius Calenus, his legate, in charge of this task (BC 1.87). He put M. Scaurus in charge of this matter (De Har Resp 43).
74
He puts the young Brutus in charge of these forces (BG 7.9).
75
In charge of the right wing he put L. Voluminius, of the left L. Scipio, of the cavalry the other legates (Livy 10.40.7).
54 ARGUMENTS OF VERBS

76
Scodrae… Gabinium praefecit, Rhizoni… C. Licinium (Livy 45.26.2)
elephantis et parti copiarum pedestrium Bomilcarem praefecit (Jug 49.1)
Provinciae Q. Cassium praeficit (BC 2.21)
cur non rebus humanis aliquos otiosos deos praeficit (De Nat Deor 3.93).
This data set illustrates the impact of pragmatic factors on word order statistics in Latin. When a person is put in
charge of something, either the person or what he is put in charge of are likely to be information already established in
the discourse. There is also an elevated incidence of contrastive status: different people get assigned different tasks in
military and political affairs. Consequently there is a high rate of topicalization. This has two consequences. First, the
relative frequencies of the DO – IO and IO – DO orders will depend to a considerable extent on which argument is
more likely to be a topic, not on grammatical factors. Second, the raw absolute frequency of the neutral order is going
to be misleading, since it will include a significant proportion of examples with string vacuous topicalization which
again are driven by pragmatic rather than grammatical factors.

Bellum inferre
A completely different distribution is evidenced by the precompiled phrase bellum inferre ‘wage war on.’ Here the order
is very regularly IO – DO in Caesar
77
ut… minus facile finitimis bellum inferre possent (BG 1.2)
uti toti Galliae bellum inferrent (BG 1.30)
neve his sociisque eorum bellum inferret. (BG 1.35)
neque his neque eorum sociis iniuria bellum inlaturum (BG 1.36)
ne aut Haeduis aut eorum sociis bellum inferret (BG 1.43)
et populo Romano bellum intulisse. (BG 2.14)
neque priores populo Romano bellum inferre (BG 4.7)
qui sibi Galliaeque bellum intulissent (BG 4.16).
In some examples the presence of a demonstrative and/or coordination may be an additional factor, but overall it is
clear that the nonreferential direct object behaves quite differently from a regular referential direct object.

76
In charge of Scodra he put Gabinius, of Rhizon C. Licinius (Livy 45.26.2). He put Bomilcar in charge of the elephants and a part of the infantry forces (Jug 49.1). He put Q.
Cassius in charge of the province (BC 2.21). Why does it not put some idle gods in charge of human affairs (De Nat Deor 3.93).
77
So that they could less easily wage war on their neighbours (BG 1.2). In order to wage war on all of Gaul (BG 1.30). That he not wage war on them and their allies (BG 1.35).
That he would not wrongfully wage war on them or their allies (BG 1.36). That he not wage war either on the Aedui or their allies (BG 1.43). That they have waged war on the
Roman people (BG 2.14). That they were not the first to wage war on the Roman people (BG 4.7). Who had waged war on himself and Gaul (BG 4.16).
 ARGUMENTS OF VERBS 55

1.3 OBLIQUE ARGUMENTS: LOCATIVES AND INSTRUMENTS

Conlocare
Conlocare ‘place in’ takes an accusative of the object whose location is changed and a prepositional phrase (or locative)
for the new location. The neutral order has the direct object preceding the location. Here are some examples from
Caesar
78
tum magni ponderis saxa et praeacutas trabes in muro conlocabant (BG 2.29)
constituit cohortes duas in Nantuatibus conlocare (BG 3.1)
C. Fabium legatum et L. Minucium Basilum cum legionibus duabus in Remis conlocat (BG 7.90)
exercitum in hibernis conlocare (BG 5.24)
rates duplices… e regione molis conlocabat (BC 1.25)
tribunal suum iuxta C. Treboni… sellam conlocavit (BC 3.20)
Q. Tullium Ciceronem et P. Sulpicium Cabilloni et Matiscone in Haeduis ad Ararim rei frumentariae causa conlocat
(BG 7.90).

78
Then they were placing stones of great weight and sharpened beams in the wall (BG 2.29). He decided to place two cohorts in territory of the Nantuates (BG 3.1). He placed
the legate C. Fabius and L. Minucius Basilus with two legions in the territory of the Remi (BG 7.90). To place the army in winter quarters (BG 5.24). He placed a pair of rafts
directly against the breakwater (BC 1.25). Placed his tribunal next to the chair of C. Trebonius (BC 3.20). He placed Q. Tullius Cicero and P. Sulpicius in Cabillonum and
Matisco in the territory of the Aedui near the Arar on account of the grain supply (BG 7.90).
56 ARGUMENTS OF VERBS

In the last example (BG 7.90) both arguments are contrastive in the paragraph context. Now we will give some
examples from Cicero
79
spem malefici praesentis in incerto reliqui temporis eventu conlocares (Pro Quinct 83)
suum praesidium in capite atque cervicibus nostris conlocare (De Leg Agr 2.74)
gentem Allobrogum in vestigiis huius urbis… conlocarent (Cat 4.12)
sedem omnium rerum ac fortunarum suarum Romae conlocavit (Pro Arch 9)
Libertatis simulacrum in ea domo conlocabas quae… (De Dom 110)
aedilitatem duobus in locis… conlocavit (De Dom 112)
colonias sic idoneis in locis contra suspicionem periculi conlocarunt ut… (De Leg Agr 2.73).
In the last example (De Leg Agr 2.73) the direct object seems to be scrambled to the left of sic. In a number of other
examples the locative argument is scrambled
80
uti in his locis legionem hiemandi causa conlocaret (BG 3.1)
loco idoneo et occulto omnem exercitum equitatumque conlocavit (BC 3.38)
tu inter eius modi mulieres praetextatum tuum filium… conlocavisti (Verr 2.5.137)
in possessione praediorum eius familiam suam conlocavit (Pro Flacc 72)
in civis… cruore et paene ossibus simulacrum non libertatis publicae, sed licentiae conlocasti (De Dom 131)
in visceribus eius qui urbem… conservasset monumentum deletae rei publicae conlocaris (De Dom 137: app. crit.)
ut in amore atque in voluptatibus adulescentiam suam conlocaret (Pro Cael 39)
in eius tetrarchia unum ex Graecis comitibus suis conlocarat (Phil 2.94).
In some of these examples demonstratives or anaphoric pronouns help to trigger the scrambling. Topicalization of
part of the object phrase (with stranding of the participial modifier in its neutral order position) occurs in the following
example
81
sacra ista nostri maiores adscita ex Phrygia Romae conlocarunt (De Har Resp 27.)

Liberare
Liberare ‘set free from’ takes a direct object and an ablative of separation. The following examples have the neutral
order with the source argument in the ablative following the theme argument in the accusative
82
qui bis Italiam obsidione et metu servitutis liberavit (Cat 4.21)
Ego Kalendis Ianuariis senatum et bonos omnis legis agrariae maximarumque largitionum metu liberavi (In Pis 4)
Syriamque immani Parthorum impetu liberavit (Phil 11.35)

79
You placed the hope of present crime in the uncertain outcome of future time (Pro Quinct 83). To place their garrison on our head and necks (De Leg Agr 2.74). So that they
might place the nation of the Allobroges on the remains of this city (Cat 4.12). He placed the seat of all his affairs and fortune in Rome (Pro Arch 9). You placed a statue of
Liberty in that house which… (De Dom 110). He placed the money for his aedileship in two places (De Dom 112). Who placed colonies in suitable places against the
suspicion of danger in such a way that… (De Leg Agr 2.73).
80
So that he might place his legion in these places to pass the winter (BG 3.1). He placed the whole army and cavalry in a suitable and hidden place (BC 3.38). You placed your
young son among women of that sort (Verr 2.5.137). Placed the slaves of his own household in possession of her estate (Pro Flacc 72). In the blood and almost the bones of a
citizen… you placed a statue, not of public liberty, but of licence (De Dom 131). In the viscera of the one who had preserved the city you placed a monument of the
destruction of the republic (De Dom 137). That he might spend his youth in love and pleasures (Pro Cael 39). He had placed one of his Greek companions in his tetrarchy
(Phil 2.94).
81
Our ancestors, having adopted them from Phrygia, placed those rites in Rome (De Har Resp 27).
82
Who twice set Italy free from siege and fear of slavery (Cat 4.21). On the first of January I set the senate and all good men free from the fear of an agrarian law and a very large
distribution of doles (In Pis 4). Set Syria free from the savage attack of the Parthians (Phil 11.35).
 ARGUMENTS OF VERBS 57

83
qui hunc populum dominatu regio liberavit (De Orat 2.225)
Et tum quidem incolumis exercitum obsidione liberavit (De Div 1.51)
agrumque Volaterranum et oppidum omni periculo in perpetuum liberavit (Ad Fam 13.4.2)
multas civitates acerbissimis tributis… liberavi (Ad Fam 15.4.2).
It is also possible for one of the arguments to get scrambled
84
remque publicam sine armis maximo civilis belli periculo liberavit (Phil 13.8)
incensione urbem… vastitate Italiam, interitu rem publicam liberavi (Pro Sull 33)
servitute Graeciam liberavisset (De Amic 42)
maximisque erroribus animos hominum liberavisse (De Fin 1.14).
In the first example the direct object has scrambled along with the prepositional phrase, so this is not an example of
neutral order. The motives for scrambling can often be quite subtle. Compare
85
qui civitatem dominatu regio liberavit (Pro Planc 60)
qui… dominatu regio rem publicam liberavit (Phil 1.13).
The neutral order means that what Brutus did was free the state from regal despotism, the scrambled order that what
Brutus did about regal despotism was free the republic from it.

Instrumental arguments
In this section we check the position of instrumental arguments for a number of verbs: cingere ‘surround with,’ complere
‘fill up with,’ cumulare ‘heap up with,’ aspergere ‘sprinkle with,’ oblinere ‘smear with.’ The neutral order has the patient
argument in the accusative preceding the instrument argument in the ablative
86
ut omnem rem publicam vestris militibus… vestris praesidiis cingeretis (De Leg Agr 2.99)
sub iugo Albae Longae castra vallo cingunt (Livy 7.39.8)

83
Who freed this people from regal despotism (De Orat 2.225). And on that occasion he set the army free from siege without harm to himself (De Div 1.51). And set the
Volaterran land and town free from all danger forever (Ad Fam 13.4.2). I set many cities free from very harsh taxation (Ad Fam 15.4.2).
84
And he set the republic free from the danger of civil war without battle (Phil 13.8). I set free the city from arson, Italy from devastation, the republic from extinction (Pro Sull
33). When he had set Greece free from slavery (De Amic 42). And to have set the souls of men free from the greatest errors (De Fin 1.14).
85
Who set the state free from regal despotism (Pro Planc 60). Who set the republic free from regal despotism (Phil 1.13).
86
That you might surround the whole republic with your soldiers… your garrisons (De Leg Agr 2.99). Under the ridge of Alba Longa surround their camp with a palisade (Livy
7.39.8).
58 ARGUMENTS OF VERBS

87
et servitia iam ferrum auro cingunt (Pliny NH 33.23)
superiorem partem collis… densissimis castris compleverant (BG 7.46)
murosque armatis compleverunt (BC 3.81)
cum terras larga luce compleverit (De Nat Deor 2.49)
virtutem militum laudibus cumulat (Tac Hist 2.57)
struem rogi nec vestibus nec odoribus cumulant (Tac Ger 27.2)
frequenti contione pietatem militum laudibus cumulat (Tac Hist 3.36)
ne aram sanguine aspergeret (De Nat Deor 3.88)
deinde conmissuras et vincula luto oblinito (Col De Arb 26.9)
oportebit palorum capita… mixto fimo cum cinere oblinere (Col 5.9.3: app.crit.)
et anum liquida pice oblinunt (Col 6.30.9)
Vitiles fimo bubulo oblinunt (Varro RR 3.16.16).
The scrambled order also occurs
88
Nervii vallo pedum ix et fossa pedum xv hiberna cingunt (BG 5.42)
pari altitudinis fastigio oppidum cingebant (BG 7.69)
pars corona vallum cingunt (Livy 4.27.7)
latifundiis vestris maria cinxistis (Sen Ep 89.20)
si rana saliva sua oculum asperserit (Pliny NH 28.73)
fimoque bubulo summam taleam oblinito (Cato 46.2).

Goal phrases
So far, as a control on the analysis, we have held the verb steady and allowed the argument phrases to vary. In the next
two sections we will change the perspective, holding the argument phrase steady and allowing the verbs to vary.

In castra, in hiberna
In our first data set, we look at two prepositional goal phrases in transitive clauses in Caesar: in castra ‘into camp,’ in
hiberna ‘into winter quarters.’ The

87
Even slaves surround their iron rings with gold (Pliny NH 33.23). They had filled up the upper part of the hill with camps crowded together (BG 7.46). And filled up their
walls with armed men (BC 3.81). When it has filled up the lands with plentiful light (De Nat Deor 2.49). Heaps praises on the courage of the soldiers (Tac Hist 2.57). They
heap up the pile of the funeral pyre with neither clothes nor perfumes (Tac Ger 27.2). In a crowded assembly he heaps praises on the loyalty of the soldiers (Tac Hist 3.36).
Lest he sprinkle the altar with blood (De Nat Deor 3.88). Then smear the joints and the ties with mud (Col De Arb 26.9). It will be necessary to smear the tops of the stakes
with dung mixed with ashes (Col 5.9.3). And smear the anus with liquid pitch (Col 6.30.9). They smear the wicker ones with cow dung (Varro RR 3.16.16).
88
The Nervii surround their winter quarters with a palisade of nine feet and a trench of fifteen feet (BG 5.42). Surrounded the town with a slope equal in elevation (BG 7.69).
Part surround the palisade with a cordon of troops (Livy 4.27.7). You have surrounded the seas with your estates (Sen Ep 89.20). If a frog has sprinkled the eye with its saliva
(Pliny NH 28.73). Smear the top of the cutting with cow dung (Cato 46.2).
 ARGUMENTS OF VERBS 59

following examples probably have broad scope focus; the goal argument follows the direct object
89
circiter meridiem exercitum in castra reduxit (BG 1.50)
copias in castra reducunt (BC 1.42)
Itaque Curio exercitum in castra reducit (BC 2.35)
frumentumque in hiberna comportavissent (BG 5.26)
legiones in hiberna mittit (BG 7.90).
The goal argument can scramble to the left of the direct object
90
rursus in hiberna legiones reduxit (BG 6.3)
maturius paulo quam tempus anni postulabat in hiberna in Sequanos exercitum deduxit (BG 1.54)
levi facto equestri proelio… in castra exercitum reduxit (BG 7.53)
ad committendum proelium alienum esse tempus arbitratus… in castra legiones reduxit (BG 4.34).
Notice that in some of these examples the scrambled phrase is focused. Like topics, scrambled phrases can have a
superimposed focus. For instance in the last two examples (BG 7.53; 4.34) the camp is contrasted with the expected
full scale battle. This focus is in addition to the regular focus projected over the verb phrase (the OV structure). In the
latter the direct object can be established or implicitly available information (legiones, castra, and hiberna in BG 6.3), but
the whole OV phrase is a weak focus.

Legatos mittere
For our second data set we will look at the phrase legatos mittere ‘to send envoys’ in finite indicative active occurrences.
Let us start with place name goal phrases; these follow the direct object in the neutral order
91
legatosque Tibur miserunt (Livy 9.30.6)
Hieronymus legatos Carthaginem misit (Livy 24.6.7)
legatos Romam miserunt (Livy 39.54.4)
Haliartii… legatos in Macedoniam miserunt (Livy 42.46.9)
legatos Athenas miserunt (Nepos 2.6.4)
Lacedaemonii legatos Athenas miserunt (Nepos 2.8.2)

89
Around noon he withdrew the army into camp (BG 1.50). Withdraw their forces into camp (BC 1.42). And so Curio withdraws the army into camp (BC 2.35). And had
brought grain into winter quarters (BG 5.26). He sends the legions into winter quarters (BG 7.90).
90
He withdrew the legions back into winter quarters (BG 6.3). A little earlier than the time of year required withdrew the army into winter quarters in the territory of the Sequani
(BG 1.54). After a light cavalry battle he withdrew the army into camp (BG 7.53). Thinking the time to be unfavourable for joining battle he withdrew the legions into camp
(BG 4.34).
91
They sent envoys to Tibur (Livy 9.30.6). Hieronymus sent envoys to Carthage (Livy 24.6.7). They sent envoys to Rome (Livy 39.54.4). The Haliartii sent envoys to Macedonia
(Livy 42.46.9). Sent envoys to Athens (Nepos 2.6.4). The Lacedaemonians sent envoys to Athens (Nepos 2.8.2).
60 ARGUMENTS OF VERBS

92
sed etiam legatos Lacedaemonem miserunt (Nepos 6.3.3)
legatos in Bithyniam miserunt (Nepos 23.2.2.)
In the following examples the goal phrase has been scrambled to the left of the direct object; in all cases the scrambled
goal phrase associates with a preceding focus particle (usque, et), creating an additional focus phrase preceding the
regular weak focus on OV
93
usque in Hispaniam legatos ac litteras misit (Pro Leg Man 9)
usque in Pamphyliam legatos deprecatoresque misissent (Pro Leg Man 35)
et in Achaiam legatos misit (Livy 32.5.4)
et ad tyrannum legatos miserunt… et Romam… legatos miserunt (Livy 35.13.3)
et Romam utraque pars miserunt legatos et inter se ipsi de reconcilianda concordia agebant (Livy 41.25.2).
Personal name (and title) goal phrases are another common type; they follow the same rules of distribution as the place
name goals
94
legatos ad Caesarem miserunt (BG 2.28; 6.32)
legatosque ad Hannibalem misit (Livy 23.33.4)
extemplo legatos ad Hieronymum misit (Livy 24.6.4)
legatos ad Hannibalem miserunt (Livy 25.13.2; 25.15.1)
legatos ad regem miserunt (Livy 33.20.2)
Aetoli… legatos ad Antiochum miserunt (Livy 36.26.1)
Indutiomarus… legatos ad Caesarem mittit (BG 5.3)
legatos ad Caesarem mittunt (BG 1.11; 5.20).
We take the neutral order to be broad scope focus. There is no rule requiring the goal phrase to be new information in
this construction: for instance, Tibur (Livy 9.30) in (66), and ad Hannibalem (Livy 23.33), ad Hieronymum (Livy 24.6) both
in (69) are all old information. The writer simply chooses to present the arguments as part of a broad scope focus
irrespective of their informational status: ‘this is what Indutiomarus did: he sent envoys to Caesar.’ Direct object
scrambling is discernible when another constituent intervenes between the object and the goal phrase

92
But they even sent envoys to Sparta (Nepos 6.3.3). Sent envoys to Bithynia (Nepos 23.2.2).
93
Sent envoys as far as Spain (Pro Leg Man 9). Sent envoys and intercessors as far as Pamphylia (Pro Leg Man 35). Sent envoys even to Achaea (Livy 32.5.4). They sent envoys
to the tyrant… and they sent envoys to Rome (Livy 35.13.3). Each party both sent envoys to Rome and treated with each other about restoring concord (Livy 41.25.2).
94
Sent envoys to Caesar (BG 2.28). And sent envoys to Hannibal (Livy 23.33.4). Immediately sent envoys to Hieronymus (Livy 24.6.4). Sent envoys to Hannibal (Livy 25.13.2).
Sent envoys to the king (Livy 33.20.2). The Aetolians sent envoys to Antiochus (Livy 36.26.1). Indutiomarus sent envoys to Caesar (BG 5.3). Sent envoys to Caesar (BG 1.11).
 ARGUMENTS OF VERBS 61

95
legatos extemplo Syracusas misit (Livy 24.29.5)
legatos gratulatum Romam misere (Livy 7.38.2).
There are also a number of examples in which the goal phrase is scrambled
96
Antiochus non civitatium modo quae circa se erant contrahebat praesidia sed ad Prusiam Bithyniae regem legatos
miserat (Livy 37.25.4)
extemplo ad Galam… legatos mittunt (Livy 24.48.14)
Ipse autem Ariovistus… Quam ob rem placuit ei ut ad Ariovistum legatos mitteret (BG 1.33–34)
clam profugit Apollonia Staberius. Illi ad Caesarem legatos mittunt (BC 3.12)
statim ad Caesarem legatos de pace miserunt (BG 4.27).
The pragmatic value of the scrambled phrase varies. Ariovistus in the third example (BG 1.34) is an established
referent, while Gala in the second example (Livy 24.48) and Prusias in the first (Livy 37.25) are newly introduced
referents (weak focus). In an interesting group of examples in Caesar the verb phrase associates with a focus attracting
phrase
97
qui uni ex Transrhenanis ad Caesarem legatos miserant (BG 4.16)
qui uni ex Gallia de pace ad Caesarem legatos numquam miserant (BG 6.5)
ultro ad Caesarem legatos miserat (BC 3.80).
‘They were the only ones to have sent envoys’. Apparently the association with focus has the effect of forcing
presuppositional material out of the OV structure (which is the constituent associated with the focusing phrase). Given
Caesar's tight constraints on postverbal objects, the only available option was scrambling. In V-bar syntax (§1.7) the
nonreferential direct object can show up in postverbal position, as in Livy 41.25 cited in (68) above and in the
following examples
98
ut ad regem mitterent legatos (Livy 36.12.3)
si quando Roman aliove quo mitterent legatos (Livy 38.30.7)
spatium petierint quo de ea re et ad Hostilium consulem et Romam mitterent legatos (Livy 43.4.9).

95
Sent envoys immediately to Syracuse (Livy 24.29.5). Sent envoys to Rome to offer congratulations (Livy 7.38.2).
96
Antiochus was not only gathering reinforcements from the states which were around him, but also sent envoys to Prusias, king of Bythinia (Livy 37.25.4). They send envoys
immediately to Gala (Livy 24.48.14). Moreover, Ariovistus himself… Therefore it seemed advisable to him to send envoys to Ariovistus (BG 1.33–34). Staberius fled secretly
from Apollonia. They sent envoys to Caesar (BC 3.12). At once sent envoys to Caesar concerning peace (BG 4.27).
97
Who were the only ones of the peoples across the Rhine to have sent envoys to Caesar (BG 4.16). And they were the only ones in Gaul never to have sent envoys to Caesar
concerning peace (BG 6.5). On their own initiative had sent envoys to Caesar (BC 3.80).
98
That they send envoys to the king (Livy 36.12.3). If at any time they sent envoys to Rome or anywhere else (Livy 38.30.7). They asked for time to send envoys both to
Hostilius, the consul, and to Rome concerning their matter (Livy 43.4.9).
62 ARGUMENTS OF VERBS

Impetum facere
Impetum facere ‘to make an attack’ is a precompiled phrase with a nonreferential object and a light verb. It takes a goal
phrase with the preposition in ‘against.’ Although the object can be scrambled (§1.6), it mostly follows the goal phrase
in Caesar
99
si in nostros fines impetum faceret (BG 1.44)
et in nostros impetum fecerunt (BG 3.28)
in hostes impetum fecerunt (BG 4.26)
Germani una in parte confertis turmis in hostes impetum fecerunt (BG 7.80)
equitatus Caesaris in cohortes impetum fecit (BC 1.70)
universi in hostes impetum fecerunt (BC 3.37)
in aversos nostros impetum fecerunt (BC 3.69)
ultro in nostros impetum faciebat (BC 3.69)
infestisque signis tanta vi in Pompei equites impetum fecerunt (BC 3.93).
We assume the prepositional goal phrase is an argument of the verbal complex and not simply adnominal: it can be
discontinuous from the object
100
Ita nostri acriter in hostes signo dato impetum fecerunt (BG 1.52).
By contrast, when impetum is a definite object phrase of a nonlight verb, it does not show this preference for preverbal
position
101
hostiumque impetum fortiter sustinerent (BG 2.21)
diutius nostrorum militum impetum hostes ferre non potuerunt (BG 4.35)
primum hostium impetum multis ultro vulneribus inlatis fortissime sustinuerint (BG 5.28)
ut… hostium impetum magno animo sustineant. (BG 7.10)
Eorum impetum Galli sustinere non potuerunt (BG 7.13).
So the normal rule that a goal phrase follows an object phrase does not apply to the light verb phrase impetum facere due
to the special properties of the indefinite object in the set phrase.

99
If he made an attack against our territory (BG 1.44). And made an attack against our forces (BG 3.28). Made an attack against the enemy (BG 4.26). The Germans in one part
of the battlefield, with their cavalry squadrons in close order, made an attack against the enemy (BG 7.80). Caesar's cavalry made an attack against the cohorts (BC 1.70). All
together made an attack against the enemy (BC 3.37). Made an attack against our forces from the rear (BC 3.63). Conversely began to make an attack against our forces (BC
3.69). With their standards raised made an attack against Pompey's cavalry with such great force (BC 3.93).
100
Our forces, when the signal was given, made an attack against the enemy so fiercely (BG 1.52).
101
That they should bravely withstand the attack of the enemy (BG 2.21). The enemy was not able to withstand any longer the attack of our soldiers (BG 4.35). They withstood
the first attack of the enemy very bravely, having, in addition, inflicted many wounds (BG 5.28). That they should courageously withstand the attack of the enemy (BG 7.10).
The Gauls were not able to withstand their attack (BG 7.13).
 ARGUMENTS OF VERBS 63

Source phrases
This data set consists of a number of common source phrases with the preposition ex, for instance ex urbe, ex castris, ex
hibernis, ex agris, ex oppido in transitive clauses. We will again start with examples that are in broad scope focus
102
Athenienses copias ex urbe eduxerunt (Nepos 1.5.2)
praesidium Lacedaemoniorum ex arce pepulerunt (Nepos 16.3.3)
quanto cum periculo legionem ex hibernis educturus esset (BG 5.47)
essedarios ex silvis emittebat (BG 5.19)
pecora atque homines ex agris in silvas compellebat (BG 5.19)
Varus praesidium… ex oppido educit (BC 1.13)
ut Gallonium ex oppido expellerent (BC 2.20)
oppida muniunt, frumentum ex agris in oppida comportant (BG 3.9)
cum praedam ex agris agerent (Livy 1.1.5)
interim Locrenses frumentum… ex agris in urbem rapere (Livy 24.1.2)
Caralitani… sua sponte Cottam ex oppido eiciunt (BC 1.30).
In all instances the source phrase follows the direct object (and precedes the goal phrase if there is one). This order is
not the result of any direct encoding of pragmatic values narrowly associated with individual constituents. It is not
simply serialization according to a principle such as “topic before focus” or “old information before new.” In a number
of examples, (for instance BG 5.47; BC 1.30; BC 2.20) both the direct object and the source phrase are old
information. The last example (BC 1.30) tells us that what the people of Cagliari did was, of their own accord, to throw
Cotta out of the city. It is not designed to tell us what the people of Cagliari did to Cotta, nor what was done to whom
out of the city nor what was done to Cotta out of the city. This serialization is therefore a properly syntactic rule of
neutral word order in broad scope focus, not a direct encoding of informational status in each individual context.
In the following examples the source phrase moves to the left of the direct object
103
Vetat ex agro culto… ullam partem sumi sepulcro (De Leg 2.67)
Ex agro homines traducis in forum, ab aratro ad subsellia (Verr 2.3.26)

102
The Athenians led their forces out of the city (Nepos 1.5.2). They drove the garrison of Lacedaemonians from the citadel (Nepos 16.3.3). With what great danger he would
lead the legion out of winter quarters (BG 5.47). He would send charioteers out of the woods (BG 5.19). Drove the herds and the people from the fields into the woods (BG
5.19). Varus leads the garrison out of the town (BC 1.13). To expel Gallonius from the town (BC 2.20). They fortify their towns, carry grain from the fields into the towns
(BG 3.9). When they were driving booty from the fields (Livy 1.1.5). Meanwhile the Locrians were bringing grain from the fields into the city (Livy 24.1.2). The Caralitani of
their own accord throw Cotta out of the town (BC 1.30).
103
He forbids that from cultivated land any part be taken for a grave (De Leg 2.67). You drag men from their farms into the forum, from the plough into the courts (Verr
2.3.26).
64 ARGUMENTS OF VERBS

104
quam subito non solum ex urbe verum etiam ex agris ingentem numerum perditorum hominum collegerat! (Cat 2.8)
ex finitimis municipiis frumentum comportare (BC 1.18).
In the first example (De Leg 2.67) the accusative is a passive subject. The first two examples (De Leg 2.67; Verr 2.3)
are strong topics, the third (Cat 2.8) is an additional focus. Object topicalization is also possible
105
Hostes, Romani, si ex agro expellere voltis (Livy 25.12.9).

1.4 ADJUNCTS
Recall that while arguments are obligatorily projected into the syntax by the head, adjuncts are syntactically optional.
Prototypically arguments represent the core participants in a situation, while adjuncts are adverbials often representing
circumstantial information. In this section we shall look at expressions of instrument, cause, time and place, manner as
well as comitatives and ablative absolutes.

Instrumentals
The distribution of instrumental adjuncts is similar to that of instrumental arguments, which was analyzed in §1.3. In
the neutral order, instrumental adjuncts appear to the right of objects
106
umerum apertum gladio appetit (BC 2.35)
Caesari gladio minitarentur (Sall Cat 49.4)
qui tabulas publicas municipi manu sua corrupisse iudicatus sit (Pro Clu 125)
postemque tremebunda manu tetigit (De Dom 134)
caput sinistra manu perfricans (In Pis 61)
spicas falcibus desecantem (Livy 42.64.3: app. crit.)
mulierem veneno interfecit (Pro Clu 31)
Habitum veneno tollere conatus sit (Pro Clu 45)
Habitum per servum medici veneno necare (Pro Clu 61)
Boudicca vitam veneno finivit (Tac Ann 14.37).

104
How quickly he had collected a huge number of desperate men not only from the city but even the fields! (Cat 2.8). To bring grain from neighbouring towns (BC 1.18).
105
If you wish, Romans, to drive the enemy from your land (Livy 25.12.9).
106
Struck at his exposed shoulder with a sword (BC 2.35). Threaten Caesar with a sword (Sall Cat 49.4). Who was judged to have falsified by his own hand the public records of
the town (Pro Clu 125). And touched the door post with a trembling hand (De Dom 134). Scratching his head with his left hand (In Pis 61). Cutting off the ears of corn with
scythes (Livy 42.64.3). Killed the woman with poison (Pro Clu 31). Tried to eliminate Habitus by poison (Pro Clu 45). To kill Habitus by poison through the agency of the
doctor's slave (Pro Clu 61). Boudicca ended her life by poison (Tac Ann 14.37).
 ARGUMENTS OF VERBS 65

The instrumental can also appear to the left of the object, adjoined to the verb phrase like a scrambled argument
107
gladio manum praecidit (De Inv 2.59)
gladio vomicam eius aperuit (De Nat Deor 3.70)
gladio nervos incidere (Livy 37.42.5)
veneno uxorem suam necare (Sen Contr 9.2.13)
Eodemque veneno C. Oppianicum fratrem necavit (Pro Clu 30)
cum ipse manu mea coniugem liberosque interfecerim (Livy 26.15.14).
In a movement framework, you could think of this as involving scrambling from the preverbal position or simply base
generation in an alternative position. While the instrumental tends to follow a direct object, it tends to precede a
locative or goal phrase
108
ibi arbores pedicino in lapide statuito (Cato 18.4)
medias vites vinclis in terram defigito (Cato 41.4)
surculos… falce acuta ex una parte deradito (Col 5.11.4; cp. De Arb 26.4)
Q. Fabium… gladio per pectus transfigit (Livy 2.46.4)
finitumos armis aut metu sub imperium suum coegere (Jug 18.12).

Ferro
We can further control the analysis by looking at the distribution of a single word. Ferro ‘by the sword’ is, predictably,
well attested. In the neutral order it follows the (in) direct object
109
vobis atque huic urbi ferro flammaque minitantem (Cat 2.1)
eorumque advocationem manibus ferro lapidibus discussisti (De Dom 54)
Comites vero Antoni… huic urbi ferro ignique minitantur (Phil 11.37)
quia Drusum ferro, Metellum veneno sustulerat (De Nat Deor 3.81)
me L. Tarquinium Superbum… ferro igni quacumque dehinc vi possim exsecuturum (Livy 1.59.1)
ut pontem ferro igni quacumque vi possint interrumpant (Livy 2.10.4)

107
Cut off his hand with a sword (De Inv 2.59). Opened his boil with his sword (De Nat Deor 3.70). To cut their tendons with a sword (Livy 37.42.5). To kill his wife by poison
(Sen Contr 9.2.13). He killed his brother, C. Oppianicus, with the same poison (Pro Clu 30). Since with my own hand I have killed my wife and children (Livy 26.15.14).
108
There set the posts in the stone with the bolt (Cato 18.4). Fix the middle of the vines to the ground with fasteners (Cato 41.4). Scrape the cuttings on one side with a sharp
pruning knife (Col 5.11.4). Stabbed Q. Fabius through the chest with a sword (Livy 2.46.4). Forced their neighbours under their rule by arms or by fear (Jug 18.12).
109
Threatening you and this city with sword and fire (Cat 2.1). You dispersed their body of advocates by fists, sword, and stones (De Dom 54). But the companions of Antony
are threatening this city with sword and fire (Phil 11.37). Because he had eliminated Drusus by the sword, Metellus by poison (De Nat Deor 3.81). That I shall pursue L.
Tarquinius Superbus with sword, with fire, and with whatever force I can from now on (Livy 1.59.1). That they break apart the bridge by sword, by fire, and by any force they
could (Livy 2.10.4).
66 ARGUMENTS OF VERBS

110
agrumque Campanum ferro ignique est depopulatus (Livy 23.46.9)
si legatos nostros ferro atque armis petierunt (Livy 26.31.3)
multos mortalis ferro aut fuga extinxit (Jug 42.4)
quinquaginta milium spatium ferro flammisque pervastat (Tac Ann 1.51)
quorum avi proavique… exercitus nostros ferro vique ceciderint (Tac Ann 11.23)
eodem ictu brachia ferro exolvunt (Tac Ann 15.63.)
It is possible that some of the direct objects have been string vacuously scrambled, since they tend to represent given
information. In any case, the instrumental can also scramble (or get adjoined) to the left of the object(s)
111
Idcircone… nobilitas armis atque ferro rem publicam reciperavit ut… (Pro Rosc Am 141)
quam facile ferro cotidianisque insidiis… Cn. Pompeium foro curiaque privarit (De Dom 67)
cum ille saxis et ignibus et ferro vastitatem meis sedibus intulisset (De Har Resp 15)
ferro flammaque omnia pervastant (Livy 35.11.12)10
emissi militum globi verberibus et intento ferro turbatos disiecere (Tac Ann 14.61)
qui ferro alios fugaret, alios domi contineret (De Har Resp 6).
In the last example (De Har Resp 6) scrambling is triggered by left node raising (if ferro is understood as modifying
both verb phrases; cp. De Har Resp 58 in (88) below). Nonreferential viam seems to resist the neutral direct object
position and consequently to remain inside the scope of the instrumental
112
ille ferro quacumque ibat viam facere (Livy 3.48.6)
ferro viam facientem (Livy 7.33.10)11
ferro viam inventuros (Tac Hist 4.20)
inter obstantis et armatos ferro viam patefecit (Tac Ann 1.32).
As already noted, goals and sources tend to follow the instrumental

110
Laid waste the Campanian territory by sword and fire (Livy 23.46.9). If they attacked our envoys with sword and arms (Livy 26.31.3). Got rid of many men by the sword or
by exile (Jug 42.4). Laid waste an area of fifty miles with sword and fire (Tac Ann 1.51). Whose grandfathers and great-grandfathers slaughtered our armies by sword and
violence (Tac Ann 11.23). With the same sword stroke they opened the veins of their arms (Tac Ann 15.63).
111
Did the nobility recover the republic by arms and the sword with the intent that… (Pro Rosc Am 141). How easily did he keep Cn. Pompeius from the forum and the curia
by the sword and daily ambushes (De Dom 67). Since he brought devastation to my residence with stones and the sword (De Har Resp 15). With sword and fire they
devastate everything (Livy 35.11.12). Bands of soldiers having been sent out scattered them in disorder with lashes and drawn sword (Tac Ann 14.61). Who with the sword
drove some away, confined others to home (De Har Resp 6).
112
He made a path with his sword wherever he went (Livy 3.48.6). Making a path with my sword (Livy 7.33.10). That they will find a path with the sword (Tac Hist 4.20).
Opened a path with the sword through opposing and armed men (Tac Ann 1.32).
 ARGUMENTS OF VERBS 67

113
Cn. Pompeium ferro domum compulit (De Har Resp 58)
tribunos plebis ferro e rostris expelleret (Pro Sest 84)
eum civem vi ferro periculis urbe omnibus patriae praesidiis depulit quem… (De Har Resp 58)
eius furores… hoc ferro et hac dextera a cervicibus vestris reppuli (Pro Mil 77)
ut primos fugientium caedant, turbam insequentium ferro et vulneribus in hostem redigant (Livy 37.43.3).

Causa
Next we come to adverbials indicating the motive, objective or cause for an action. Here we will consider the
distribution of causa-phrases in Caesar. We find pretty much the same options as already established for instrumental
adverbials. One common position for causa-phrases is immediately preceding the verb
114
quas… in Apulia hibernorum causa disposuerat (BC 1.14)
equitatumque omnibus locis iniciendi timoris causa ostentare coeperunt (BG 7.55)
in alteram partem item cohortandi causa profectus (BG 2.21)
ut in his locis legionem hiemandi causa conlocaret (BG 3.1)
praefectos tribunosque militum complures in finitimas civitates frumenti commeatusque petendi causa dimisit (BG
3.7: app. crit.)
praesidiumque cohortium duodecim pontis tuendi causa ponit (BG 6.29)
iii cohortes Orici oppidi tuendi causa reliquit (BG 3.39)
Menapii legatos ad eum pacis petendae causa mittunt (BG 6.6)
Q. Tullium Ciceronem… Cavilloni… rei frumentariae causa conlocat (BG 7.90).
Another common position is adjoined to the verb phrase

113
He drove Cn. Pompeius to his house by the sword (De Har Resp 58). In order to drive the tribunes of the plebs from the Rostra with the sword (Pro Sest 84). By violence,
the sword, and peril he drove that citizen from the city, from every protection afforded by the fatherland, that citizen whom… (De Har Resp 58). I have driven off his fury
from your necks by this sword and by this right hand (Pro Mil 77). That they kill the first of those fleeing, that with the sword and wounds drive the mob of those following
against the enemy (Livy 37.43.3).
114
Which he had stationed in Apulia for winter quarters (BC 1.14). They began to display their cavalry in all locations for the sake of striking fear (BG 7.55). He set off in the
other direction for the sake of similarly exhorting (BG 2.21). That he station his legion in these places to pass the winter (BG 3.1). He sent out into the neighbouring states
several prefects and military tribunes for the sake of finding grain and supplies (BG 3.7). He stationed a garrison of twelve cohorts for the purpose of guarding the bridge (BG
6.29). Left three cohorts at Oricum for the purpose of guarding the town (BG 3.39). The Menapians send envoys to him for the sake of seeking peace (BG 6.6). He stationed
Q. Tullius Cicero at Cavillonum for the sake of the grain supply (BG 7.90).
68 ARGUMENTS OF VERBS

115
ut communis libertatis causa arma capiant (BG 7.4)
praesentis periculi atque inopiae vitandae causa omnem orationem instituisse (BC 3.17)
neque recusandi… causa legatos ad Caesarem mittere audebant (BG 5.6)
qui praedae ac belli inferendi causa ex Belgio transierant (BG 5.12)
cum Caesar pabulandi causa tres legiones atque omnem equitatum cum C. Trebonio legato misisset (BG 5.17)
sui purgandi causa ad eum legatos mittunt (BG 6.9)
auxilii petendi causa Romam ad senatum profectus (BG 6.12)
stipendii augendi causa regis domum obsidere (BC 3.110).
If there are other adverbials or scrambled phrases, the causa-phrase tends to be the last in the string
116
cognoverat enim magnam partem equitatus ab iis aliquot diebus ante praedandi frumentandique causa ad
Ambivaritos trans Mosam missam (BG 4.9)
neminem postea belli inferendi causa in Britanniam transiturum confidebant (BG 4.30)
legatosque deprecandi causa ad Caesarem mittunt (BG 6.4)
ex urbe amicitiae causa Caesarem secuti (BG 1.39)
in Epirum rei frumentariae causa Q. Tillium et L. Canuleium legatum misit (BC 3.42).
The causa-phrase appears after a scrambled object or directional phrase. This observation is important. It seems that
while the order of scrambled arguments is relatively free, they still have a default order, which is the same as the default
order within the verb phrase. Causa-adjuncts like to go to the end in both strings. For instance they can cross over the
direct object once but preferably not twice.

115
That for the sake of their common freedom they take up arms (BG 7.4). That he had made his whole speech because of the present danger and with the purpose of avoiding
destitution (BC 3.17). They did not dare to send envoys to Caesar for the purpose of refusing (BG 5.6). Who had gone across from Belgium for the sake of booty and waging
war (BG 5.12). When for the sake of foraging Caesar had sent three legions and the whole cavalry with C. Trebonius, his legate (BG 5.17). For the sake of clearing themselves
they send envoys to him (BG 6.9). For the sake of seeking aid set out for the Senate at Rome (BG 6.12). For the purpose of increasing their pay besieged the king's palace (BC
3.110).
116
For he knew that a large part of the cavalry had been sent by them a few days before, for the sake of pillaging and foraging, to the Ambivariti across the Meuse (BG 4.9).
They believed that no one thereafter would cross over to Britain for the sake of waging war (BG 4.30). They send envoys for the sake of begging his pardon to Caesar (BG
6.4). Followed Caesar from the city for the sake of his friendship (BG 1.39). Sent Q. Tillius and L. Canuleius to Epirus for the sake of the grain supply (BC 3.42).
 ARGUMENTS OF VERBS 69

Time and place adverbials

(Proxima) nocte, noctu

These adverbials are well attested in preverbal position
117
multitudo… in oppidum proxima nocte convenit (BG 2.12)
cum ego ad Heracleam noctu accederem (Verr 2.5.129)
cum ad Hannibalem noctu transisset (Livy 21.12.4)
cum e carcere noctu effugisset (Livy 37.46.5)
ex oppido nocte profugit (Jug 76.1)
signa sacra noctu frangere putaretur (De Orat 2.253)
furem noctu liceat occidere et luce si se telo defendat (Pro Tull 47)
si illum ad urbem noctu accessurum sciebat (Pro Mil 49).
In the last two examples (Pro Tull 47; Pro Mil 49) noctu has strong focus, but in the others the adverbial is a simple
weak focus. In the following examples (de) nocte remains in situ when another argument is scrambled
118
in comitio… in comitium Milo de nocte venit (Ad Att 4.3.4)
dum ad Antium haec geruntur, interim Aequi… arcem Tusculanam improviso nocte capiunt (Livy 3.23.1)
magnum numerum levis armaturae et sagittariorum… noctu in scaphas et naves actuarias imponit (BC 3.62).
We also find these adverbials adjoined to the verb phrase
119
pauci… nocte fuga salutem petierunt (BC 3.97)
noctu insidias equitum conlocavit (BC 3.37)
noctu duos conservos… occidit (Pro Clu 179)
Numida… noctu Iugurthae milites introducit (Jug 12.4).
The branching phrase proxima nocte occurs with clausal scope, which is appropriate for an event modifier

117
The crowd assembled the next night in the town (BG 2.12). When I was approaching Heraclea at night (Verr 2.5.129). When he had gone over to Hannibal in the night (Livy
21.12.4). When he had escaped from prison by night (Livy 37.46.5). Fled from the town by night (Jug 76.1). Was thought to be breaking holy statues by night (De Orat 2.253).
Which permits one to kill a thief at night and during the day if he defends himself with a weapon (Pro Tull 47). If he knew he would be approaching the city at night (Pro Mil
49).
118
In the Comitium… Milo went to the Comitium just at dawn (Ad Att 4.3.4). While these things are taking place at Antium, at the same time the Aequi capture the Tusculan
citadel by surprise at night (Livy 3.23.1). He embarks a large number of light-armed troops and archers at night in skiffs and fast ships (BC 3.62).
119
A few sought safety in flight by night (BC 3.97). He laid a cavalry ambush by night (BC 3.37). At night he killed two fellow slaves (Pro Clu 179). Let in Jugurtha's soldiers by
night (Jug 12.4).
70 ARGUMENTS OF VERBS

120
quin proxima nocte Sabinus clam ex castris exercitum educat (BG 3.18)
Proxima nocte centuriones Marsi duo ex castris Curionis cum manipularibus suis xxii ad Attium Varum perfugiunt
(BC 2.27).

Romae
The following examples have the locative in its neutral position preceding the verb
121
cum sceleris sui socios… Romae reliquisset (Cat 3.3)
Fundum Cymaeum Romae mercatus est (Pro Flacc 46)
eum poenas Romae daturum (Verr 2.1.82)
duo templa se Romae dedicaturum voverat (Verr 2.4.123)
si consulem Romae habuissemus (Ad Fam 10.10.1)
consulatum reliquum Romae peragere (Ad Att 8.15a.2).
The locative in this position may have strong focus as probably in the last example (Ad Att 8.15a) or regular weak
focus as in the other examples. Scrambling of another phrase can strand the locative in the verb phrase
122
cum cc talenta tibi Apolloniatae Romae dedissent (In Pis 86)
and weak focus on another constituent, such as the direct object or an adverbial, can trigger scrambling of the locative
itself
123
Aris autem… continuo Romae matrem illam Bostaris duxit uxorem (Pro Scaur 12)
Itaque nunc Romae omnia negotia Lentuli procuro (Ad Att 9.7b.2)
Alfenus interea Romae cum isto gladiatore vetulo cotidie pugnabat (Pro Quinct 29).
If the locative is a contrastive topic, it scopes over its domain
124
qui et domi suae cum primis honestus existimatus est et Romae argentariam non ignobilem fecit (Pro Caec 10)

120
That on the next night Sabinus was secretly to lead the army out of the camp (BG 3.18). On the next night two Marsian centurions from Curio's camp along with twenty-two
common soldiers of theirs desert to Attius Varus (BC 2.27).
121
Since he left accomplices in his crime in Rome (Cat 3.3). At Rome he bought a Cymaean estate (Pro Flacc 46). That he would pay the penalty at Rome (Verr 2.1.82). Had
vowed that he would dedicate two temples at Rome (Verr 2.4.123). If we had had a consul at Rome (Ad Fam 10.10.1). To finish the rest of his consulate at Rome (At Att
8.15a.2).
122
When the Apollonians had given you two hundred talents at Rome (In Pis 86).
123
Aris, moreover, immediately married the aforementioned mother of Bostar at Rome (Pro Scaur 12). Therefore I am now attending to all of Lentulus's business at Rome (Ad
Att 9.7b.2). Meanwhile at Rome Alfenus was fighting daily with that aging gladiator (Pro Quinct 29).
124
Who in his own native land was esteemed as honorable among the first citizens and at Rome carried on a renowned banking business (Pro Caec 10).
 ARGUMENTS OF VERBS 71

125
qui Romae tribunatum pl. peteret cum in Sicilia hereditatem se petere dictitasset (Ad Att 2.1.5).
Finally, the locative can have clausal scope
126
si Romae socii incepta patravissent (Sall Cat 56.4)
Interim Romae C. Mamilius Limetanus… rogationem ad populum promulgat (Jug 40.1)
Et Romae senatus de provinciis consultus Numidiam Metello decreverat (Jug 62.10).
The first example gives the location of a complete event including the subject, the last two examples (Jug 40.1; 62.10)
shift the story scene to Rome from elsewhere.

Manner and means phrases

Virtute
Virtute, by itself or plus modifier, normally appears after the complement in preverbal position like instruments
127
summam spem civium… incredibili virtute superavit (De Amic 11)
ut turpitudinem fugae virtute delerent (BG 2.27)
eorum consilia sua virtute confirmavissent (BC 2.21)
regem Persen vi ac virtute superavit (Verr 2.1.55)
dignitatem nostram… non copiis sed virtute tueamur (Verr 2.3.9)
ita vivit ut nobilitatis dignitatem virtute tueri posse videatur (Pro Clu 111)
memoriam prope intermortuam generis sua virtute renovare (Pro Mur 16)
pacem maritimam summa virtute atque incredibili celeritate confecit (Pro Flacc 29)
impetum improborum virtute sedavit (Pro Sest 62)
Catilinam… vestro studio et virtute fregistis (Phil 4.15).
The means phrase can also be scrambled across the direct object

125
Who at Rome was seeking the tribunate of the plebs, while in Sicily was repeatedly saying he was seeking an inheritance (Ad Att 2.1.5).
126
If at Rome his allies accomplished their plans (Sall Cat 56.4). Meanwhile at Rome C. Mamilius Limetanus proposed a law to the people (Jug 40.1). And at Rome, the senate,
when deliberating about the provinces, decreed Numidia for Metellus (Jug 62.10).
127
Surpassed the highest hopes of the citizens with incredible virtue (De Amic 11). That they wipe out the disgrace of their flight by valour (BG 2.27). Confirmed the resolutions
of those by their own valour (BC 2.21). Defeated King Perses by his strength and valour (Verr 2.1.55). That we may preserve our dignity not by our wealth but by our virtue
(Verr 2.3.9). So lives that he seems able to preserve the dignity of his noble rank by means of virtue (Pro Clu 111). To renew by his own virtue the almost extinct memory of
his family (Pro Mur 16). Established maritime peace with the highest valour and incredible speed (Pro Flacc 29). Restrained the assault of the disloyal by his valour (Pro Sest
62). You broke Catiline by your zeal and valour (Phil 4.15).
72 ARGUMENTS OF VERBS

128
quo proelio legio Martia admirabili incredibilique virtute libertatem populi Romani defenderit (Phil 14.36)
quam L. Lucullus virtute… summis obsidionis periculis liberavit (Pro Leg Man 20)
et qui nobiles estis… et qui ingenio ac virtute nobilitatem potestis consequi (Pro Sest 136)
virtute, consilio, providentia mea res publica maximis periculis sit liberata (Cat 3.14).

Magnis itineribus
The manner phrase magnis itineribus is quite common with verbs of motion (for instance contendo) and a goal (or source)
argument
129
magnis itineribus ab Arimino adversus Gallos Cremonam tum obsidentes profectus (Livy 31.21.2).
The order just illustrated, with the adjunct preceding the directional arguments, is the most common and presumably
the neutral order. This order is not simply a reflex of the pragmatic values in each sentence. The goal argument can be
new information
130
reliquos Catilina per montis asperos magnis itineribus in agrum Pistoriensem abducit (Sall Cat 57.1)
magnisque itineribus ad Mevaniam, ubi tum copiae Umbrorum erant, perrexit (Livy 9.41.13)
magnis itineribus Apolloniam contendit (Livy 29.12.4)
magnis itineribus Zamam contendit. Zama quinque dierum iter ab Carthagine abest (Livy 30.29.1)
per Macedoniam magnis itineribus in iugum montium quod super Gonnos est pervenit. Oppidum Gonni viginti
milia ab Larisa abest. (Livy 36.10.10),
but it can equally well be old information

128
In which battle the Martian legion defended the liberty of the Roman people with admirable and incredible valour (Phil 14.36). Which L. Lucullus freed by his valour from the
greatest dangers of the siege (Pro Leg Man 20). Both you who are nobles and you who can attain noble rank by talent and virtue (Pro Sest 136). The republic has been freed
from the greatest dangers by my valour, judgment, and foresight (Cat 3.14).
129
Set out by forced marches from Ariminum against the Gauls then besieging Cremona (Livy 31.21.2).
130
Catiline led the rest away through rough mountains by forced marches into the land of Pistoria (Sall Cat 57.1). He proceeded by forced marches to Mevania, where the forces
of the Umbrians were then (Livy 9.41.13). Hastened by forced marches to Apollonia (Livy 29.12.4). Hastened by forced marches to Zama. Zama is a five day journey from
Carthage (Livy 30.29.1). Arrived through Macedonia by forced marches at the ridge of mountains which is above Gonni. The town of Gonni is twenty miles from Larisa (Livy
36.10.10).
 ARGUMENTS OF VERBS 73

131
si nova manus Sueborum cum veteribus copiis Ariovisti sese coniunxisset… Itaque… magnis itineribus ad
Ariovistum contendit (BG 1.37)
cum Antonio… magnis itineribus ad Antonium contendit (BC 3.30)
omnia ab Herdonea… delata… magnis itineribus ad Herdoneam contendit (Livy 27.1.6)
concilium Aetolis Heracleam indictum… magnis itineribus Heracleam duxit (Livy 28.5.13).
This nonuniformity in the pragmatic status of the goal argument indicates that the verb phrase to which the adjunct
phrase is adjoined is a broad scope focus (which naturally can include some old information). In a few instances the
neutral order is disturbed and the goal argument is scrambled, stranding the adjunct in preverbal position
132
ipse in Italiam magnis itineribus contendit (BG 1.10)
cum his ad Domitium Ahenobarbum Corfinium magnis itineribus pervenit (BC 1.15)
ad urbem ex Etruria magnis itineribus pergit (Livy 9.41.10)
Canusium magnis itineribus contendit (Livy 22.57.8)
in castra Romana magnis itineribus contendit (Livy 32.9.7).
The main focus in these examples falls on the adjunct in preverbal position. Contrast the following
133
quam maximis potest itineribus in Galliam ulteriorem contendit (BG 1.7)
quam maximis potest itineribus Viennam pervenit (BG 7.9)
quam potui maximis itineribus ad Amanum exercitum duxi (Ad Fam 15.4.7)
et consul quidem quantis maximis itineribus poterat ad collegam ducebat (Livy 27.43.12).
Here the focus is in a separate clause, which is adjoined to the main clause verb phrase.

131
If a new band of Suebi joined the veteran forces of Ariovistus… Therefore he hastened by forced marches to Ariovistus (BG 1.37). With Antony… hastened by forced
marches to Antony (BC 3.30). All these things were reported from Herdonea… hastened by forced marches to Herdonea (Livy 27.1.6). That a council had been declared for
the Aetolians at Heraclea… led his troops by forced marches to Heraclea (Livy 28.5.13).
132
He himself hastens by forced marches to Italy (BG 1.10). With these he reaches Domitius Ahenobarbus at Corfinium (BC 1.15). Proceeds to the city from Etruria by forced
marches (Livy 9.41.10). Hastens to Canusium by forced marches (Livy 22.57.8). Hastened to the Roman camp by forced marches (Livy 32.9.7).
133
Hastens to Further Gaul by the longest marches he can make (BG 1.7). Reached Vienna by the longest forced marches he can make (BG 7.9). I led my army to Amanus by
the longest forced marches I could make (Ad Fam 15.4.7). The consul, in fact, was leading his forces to his colleague by the longest forced marches he could make (Livy
27.43.12).
74 ARGUMENTS OF VERBS

Sine
In this subsection we will look at manner phrases with the preposition sine in Caesar. These can follow the direct object
134
se suaque omnia sine mora dediderunt (BG 2.15)
ut legatos sibi ad Pompeium sine periculo mittere liceret (BC 3.17)
neque se praesidium ubi constitutus esset sine auxilio Scipionis tenere posse (BC 3.36)
quod Italiam sine aliquo vulnere cepissent, quod duas Hispanias… pacavissent (BC 3.73)
se… neque exercitum sine magno commeatu atque molimento in unum locum contrahere posse (BG 1.34).
In the last example a goal follows the manner phrase in the now familiar pattern. However it is more usual for the
manner phrase to be adjoined to the verb phrase or to the whole clause
135
sine ullo maleficio iter per provinciam facere, propterea quod aliud iter haberent nullum (BG 1.7)
sine ullo periculo tantam eorum multitudinem nostri interfecerunt (BG 2.11)
sine metu trans Rhenum in suos vicos remigraverant (BG 4.4)
sine ullo vulnere victoria potiri (BG 3.24)
sine imperio tantas copias reliquisset (BG 7.20: object oriented)
ut… ipsi sine periculo ac timore Hiberum copias traducerent (BC 1.65)
sine pugna et sine volnere suorum rem conficere posse (BC 1.72)
sine volnere tantas res confecisse videbantur (BC 1.74)
sine periculo legionum et paene sine vulnere bellum conficiemus (BC 3.86)
sine impedimentis Caesar legiones transportaverat (BG 4.30).
The focus on the manner phrase is usually in addition to that on the verb phrase. What they wanted to do was to get
their troops across the Ebro and to

134
Surrendered themselves and all their property without delay (BG 2.15). That he be permitted to send envoys to Pompey without danger (BC 3.17). And that he could not
hold the stronghold where he had been stationed without aid from Scipio (BC 3.36). Because they had seized Italy without any injury, because they had pacified the two Spains
(BC 3.73). And that he could not concentrate his army in one place without great supplies and labour (BG 1.34).
135
To march through the province without causing any injury, because they had no other route (BG 1.7). Without any danger our men killed such a great multitude of them (BG
2.11). They had returned without fear across the Rhine to their own villages (BG 4.4). To gain victory without any wound (BG 3.24). Had left such great forces without
command (BG 7.20). So that they might lead their forces across the Ebro without danger or fear (BC 1.65). That he could finish the business without a fight or injury (BC
1.72). Seemed to have accomplished such great things without injury (BC 1.74). We shall finish the war without danger to the legions and almost without a wound (BC 3.86).
Caesar had brought the legions across without baggage (BG 4.30).
 ARGUMENTS OF VERBS 75

do so without danger or fear (BC 1.65). Where they wanted to march was through the province, since no other route
was available to them, and they would do so without causing any damage (BG 2.11). In the last example (BG 4.30) the
manner phrase is adjoined to the whole clause which is presuppositional: the way Caesar had brought his legions across
was without baggage. (This type of focus is analyzed further in Chapter 3.)

Comitatives
The evidence for comitative adjuncts is a bit ambivalent to the extent that comitative phrases can be adnominal as well
as adverbial. Compare the following
136
Ibi turres cum ternis tabulatis erigebat (BC 1.26)
Caesar ad portum Itium cum legionibus pervenit (BG 5.5).
Subject oriented comitatives generally scope over the verb phrase including directional phrases and direct objects. This
rule applies whether the subject is overt or null (pro-dropped)
137
Roscius cum L. Caesare Capuam pervenit (BC 1.10)
M. Octavius cum iis quas habebat navibus Salonas pervenit (BC 3.9)
eodem die cum legionibus in Senones proficiscitur (BG 6.3)
ut cum legionibus quam primum Gadis contenderet (BC 2.20)
Dumnorix cum equitibus Haeduorum a castris insciente Caesare domum discedere coepit (BG 5.7)
Suebos omnes… cum omnibus suis sociorumque copiis… penitus ad extremos fines se recepisse (BG 6.10)
postero die Petreius cum paucis equitibus occulte ad exploranda loca proficiscitur (BC 1.66)
Atuatuci… cum omnibus copiis auxilio Nerviis venirent (BG 2.29)
Varro cum iis quas habebat legionibus omnem ulteriorem Hispaniam tueatur (BC 1.38)
M'.Acilium consulem iam cum legionibus mare traiecisse (Livy 36.12.10)
ipse equites impedimentaque prae se habens cum legionibus agmen cogit (Livy 44.4.12)

136
There he erected towers with three floors (BC 1.26). Caesar reaches the Itian port with his legions (BG 5.5).
137
Roscius arrives at Capua with L. Caesar (BC 1.10). M. Octavius arrives at Salonae with those ships which he had (BC 3.9). On the same day he sets out with the legions
against the Senones (BG 6.3). That he might hasten as soon as possible with his legions to Gades (BC 2.20). Dumnorix with the cavalry of the Haedui began to depart,
unknown to Caesar, from his camp for home (BG 5.7). That all the Suebi had withdrawn with all their own and their allies' forces far into their outermost territory (BG 6.10).
On the next day Petreius with a few cavalrymen secretly sets out to reconnoiter the area (BC 1.66). The Atuatuci with all these forces were coming to the aid of the Nervi (BG
2.29). Varro with those legions which he had should protect all of Further Spain (BC 1.38). That the consul M'. Acilius with his legions had already crossed the sea (Livy
36.12.10). He himself, having the cavalry and baggage in front of him, brings up the rear with his legions (Livy 44.4.12).
76 ARGUMENTS OF VERBS

138
equites nostri cum funditoribus sagittariisque flumen transgressi (BG 2.19)
Tenctheri magna cum multitudine hominum flumen Rhenum transierunt (BG 4.1)
mediam aciem Scipio cum legionibus Syriacis tenebat (BC 3.88).
In the last example (BC 3.83) the direct object is contrastive and topicalized. Object oriented comitatives generally are
placed after the object but before a locative or goal argument and before some other adjuncts
139
Q. Titurium Sabinum legatum cum legionibus tribus in Unellos, Coriosolitas Lexoviosque mittit (BG 3.11)
C. Fabium legatum cum legionibus duabus castris praesidio relinquit (BG 7.40)
C. Fabium legatum et L. Minucium Basilum cum legionibus duabus in Remis conlocat (BG 7.90)
primoque impetu unam ex his quadriremem cum remigibus defensoribusque suis ceperunt (BC 3.24)
Caesar Fabium cum sua legione remittit in hiberna (BG 5.53)
Lucterium Cadurcum, summae hominem audaciae, cum parte copiarum in Rutenos mittit (BG 7.5).
A directional or locative can be scrambled to the left of the comitative
140
Ariminum cum ea legione proficiscitur (BC 1.8)
Brundisium cum legionibus vi pervenit (BC 1.25)
uti Petreius ex Lusitania per Vettones cum omnibus copiis ad Afranium proficiscatur (BC 1.38)
M. Favonium ad flumen Haliacmonem… cum cohortibus viii praesidio impedimentis legionum reliquit (BC 3.36)
qui in Graeciam cum legionibus missus est (Ad Fam 7.30.3)

138
Our cavalry having crossed the river with the slingers and archers (BG 2.19). The Tenctheri with a great multitude of men crossed the River Rhine (BG 4.1). Scipio was
holding the middle of the battle line with the Syrian legions (BC 3.88).
139
He sends Q. Titurius Sabinus, his legate, with three legions to the territory of the Unelli, the Coriosolites, and the Lexovii (BG 3.11). He leaves C. Fabius, his legate, with two
legions as garrison for the camp (BG 7.40). He stations C. Fabius, his legate, and L. Minucius Basilus with two legions in the territory of the Remi (BG 7.90). And in the first
attack they captured one of these quadriremes with its rowers and defenders (BC 3.24). Caesar sends Fabius with his own legion back into winter quarters (BG 5.53). He
sends Lucterius Cadurcus, a man of the greatest boldness, with part of his forces into the territory of the Ruteni (BG 7.5).
140
He sets out with that legion for Ariminum (BC 1.8). He arrives at Brindisium with six legions (BC 1.25). That Petreius should set out from Lusitania through the territory of
the Vettones with all his forces to Afranius (BC 1.38). He leaves M. Favonius at the river Haliacmon with eight cohorts as a garrison for the baggage of the legions (BC 3.36).
Who has been sent to Greece with the legions (Ad Fam 7.30.3).
 ARGUMENTS OF VERBS 77

141
eae naves ad proximum portum in Histriae fines cum onerariis et magno commeatu missae (Livy 41.1.4)
L. Manlius praetor Alba cum cohortibus sex profugit, Rutilius Lupus praetor Tarracina cum tribus (BC 1.24).
or higher still
142
Itaque ab Arimino M. Antonium cum cohortibus v Arretium mittit (BC 1.11)
ad eas munitiones Caesar Lentulum Marcellinum quaestorem cum legione viiii positum habebat (BC 3.62)
ex his locis Cassius cum classe discessit (BC 3.101).
The inverse ordering in which the comitative is left peripheral also occurs
143
prius… quam cum legionibus novis consul venisset (Livy 32.8.3)
una cum reliqua Gallia Haeduis libertatem sint erepturi (BG 1.17).

Ablative absolutes
Ablative absolutes are a type of circumstantial adjunct. Since they contain an independent verbal predication, they are
strictly speaking outside the scope of this book, which is restricted to the simple sentence. However we include a brief
analysis, since the results are theoretically important. Ablative absolutes can scope over the clause
144
Germanico bello confecto… Caesar statuit sibi Rhenum esse transeundum (BG 4.16)
or over the predicate
145
Scipio digressu exercituum ab Dyrrachio cognito Larisam legiones adduxerat (BC 3.80)
or they may appear to the right of the direct object

141
These ships were sent to the nearest port in the territory of Histria with cargo vessels and a great quantity of provisions (Livy 41.1.4). L. Manilius, the praetor, flees from Alba
with six cohorts, Rutilius Lupus the praetor from Tarracina with three (BC 1.24).
142
And so he sends M. Antonius from Ariminum with five cohorts to Arretium (BC 1.11). At these fortifications Caesar had Lentulus Marcellinus, his quaestor, posted with the
ninth legion (BC 3.62). From these places Cassius departed with his fleet (BC 3.101).
143
Before the consul had arrived with new legions (Livy 32.8.3). That along with the rest of Gaul they were about to snatch away from the Aedui their freedom (BG 1.17).
144
The German war having been finished… Caesar decided that he had to cross the Rhine (BG 4.16).
145
Scipio, the departure of the armies from Dyrrachium having become known, had led his legions to Larisa (BC 3.80).
78 ARGUMENTS OF VERBS

146
quod montem gladiis destrictis ascendissent (BC 1.47)
quod provinciam Siciliam petendi causa non consulto senatu… reliquisset (Livy 37.47.6)
agrum late nullo ferme obvio armato vastavit (Livy 27.29.7)
agrum pluribus locis expositis per litora armatis late vastavit (Livy 44.10.5)
Domitius ad Pompeium in Apuliam peritos regionum magno proposito praemio cum litteris mittit (BC 1.17).
The direct object, in the last example (BC 1.17) along with the goal arguments, has probably been scrambled to the left
of a predicate scope ablative absolute.

Signo dato
Consider the following array for the ablative absolute signo dato ‘the signal having been given.’ First to the left of the
subject
147
prima luce signo dato multitudo omnis in foro instruitur (Livy 5.43.2)
hora fere nona quasi signo dato Clodiani nostros consputare coeperunt (Ad Qfr 2.3.2)
cum ad arma signo dato milites concurrissent (Livy 27.3.5)
Romam uno tempore quasi signo dato Italia tota convenit (In Pis 34).
In the last two examples (Livy 27.3; In Pis 34) the goal phrase has been scrambled to the left of an unaccusative
subject. Next predicate scope
148
Antonianae scaphae signo dato se in hostes incitaverunt (BC 3.24)
Iam montani signo dato ex castellis ad stationem solitam conveniebant (Livy 21.33.2)
Dein repente signo dato hostes invadit (Jug 50.3).
Finally, internal position is quite common with this phrase
149
ita nostri acriter in hostes signo dato impetum fecerunt (BG 1.52)

146
Because they had climbed the hill with swords drawn (BC 1.47). Because he had left his province of Sicily for the sake of running for election, without the senate having been
consulted (Livy 37.47.6). Plundered the land extensively with almost no armed resistance (Livy 27.29.7). Plundered the land extensively, armed men having been disembarked
over the shore in a number of places (Livy 44.10.5). Having offered them a great reward, Domitius sends men experienced in the districts with letters to Pompey (BC 1