1 Gadet Francoise Pecheux Michel Lingua Inatingivel Discurso Historia Linguistica Campinas Pontes
1 Gadet Francoise Pecheux Michel Lingua Inatingivel Discurso Historia Linguistica Campinas Pontes
1 Gadet Francoise Pecheux Michel Lingua Inatingivel Discurso Historia Linguistica Campinas Pontes
RECENT TITLES
Diagnosing Syntax
edited by Lisa Lai-Shen Cheng and Norbert Corver
Pseudogapping and Ellipsis
by Kirsten Gengel
Syntax and its Limits
edited by Raffaella Folli, Christina Sevdali, and Robert Truswell
Phrase Structure and Argument Structure
A Case Study of the Syntax-Semantics Interface
by Terje Lohndal
Edges in Syntax
Scrambling and Cyclic Linearization
by Heejeong Ko
The Syntax of Roots and the Roots of Syntax
edited by Artemis Alexiadou, Hagit Borer, and Florian Schäfer
Causation in Grammatical Structures
edited by Bridget Copley and Fabienne Martin
Continuations and Natural Language
by Chris Barker and Chung-chieh Shan
The Semantics of Evaluativity
by Jessica Rett
External Arguments in Transitivity Alternations
by Artemis Alexiadou, Elena Anagnostopoulou, and Florian Schäfer
Control and Restructuring
by Thomas Grano
The Interaction of Focus, Givenness, and Prosody
A Study of Italian Clause Structure
by Vieri Samek-Lodovici
The Morphosyntax of Gender
by Ruth Kramer
The Morphosyntax of Imperatives
by Daniela Isac
Sentence and Discourse
edited by Jacqueline Guéron
Optimality-Theoretic Syntax, Semantics, and Pragmatics
From Uni- to Bidirectional Optimization
edited by Géraldine Legendre, Michael T. Putnam, Henriëtte de Swart, and Erin Zaroukian
The Morphosyntax of Transitions
A Case Study in Latin and Other Languages
by Víctor Acedo-Matellán
For a complete list of titles published and in preparation for the series, see pp .
The Morphosyntax
of Transitions
A Case Study in Latin and Other Languages
VÍCTOR ACEDO-MATELLÁN
1
3
Great Clarendon Street, Oxford, OX DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Víctor Acedo-Matellán
The moral rights of the author have been asserted
First Edition published in
Impression:
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, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
Madison Avenue, New York, NY , United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number:
ISBN –––– (Hbk)
–––– (Pbk)
Printed in Great Britain by
Clays Ltd, St Ives plc
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
Contents
General preface ix
Acknowledgements x
List of abbreviations xi
Introduction
. Aim and proposal
. Methodology
.. The advantages of a theoretical approach to the
grammar of unspoken languages
.. Data and corpus
. Structure
A neo-constructionist perspective on argument structure
. Endo-skeletal versus exo-skeletal approaches to the lexicon-
syntax interface
. Three neo-constructionist theories and a predecessor
.. Hale and Keyser’s theory of lexical syntax
.. Mateu ()
.. Borer (b)
.. Distributed Morphology
. The present framework
.. Argument structure is syntax
... No l-/s-syntax distinction
... Relational and non-relational elements
... Argument structure configurations
... Adjunction of roots to functional heads
.. The semantics of argument structure: a localist-aspectual
approach
... Structural and encyclopaedic semantics
... Interpretation of functional heads and arguments
... Against root ontologies
... Aspect and argument structure
... Locality domains for special meaning
. Summary
vi Contents
imp imperative
interr interrogative
inf infinitive
instr instrumental
int internal (prefix)
ipfv imperfective
LA Locative Alternation
loc locative
m masculine
mid middle voice
n neuter
neg negation
nom nominative
num number
opt optative
part particle
partve partitive
pass passive
pfv perfective
pl plural
pluprf pluperfect
prf perfect
prs present
pst past
ptcp participle
refl reflexive
sbjv subjunctive
s-framed satellite-framed
sg singular
SI Secondary Imperfective
superl superlative
sup supine (a nominal form of the Latin verb)
th thematic vowel
transl translative
UOC Unselected Object Construction
v-framed verb-framed
voc vocative
1
Introduction
The distinction lies in the way that the head encoding transition within the vP—
Path—receives exponence during Vocabulary Insertion. In the case of v-framed
languages, Path can only receive an exponent (∅) when appearing as strictly left-
adjacent to the v head, as a prefix. This rules out any intervening material between
Path and v, accounting for the effect that these two heads are always lexicalized as a
portmanteau morph in these languages. On the other hand, in s-framed languages
this strict adjacency requirement does not exist for Path, so v and Path are free to be
phonologically realized independently from each other. Finally, I propose a refine-
ment of Talmy’s typology within the class of s-framed languages. First, there are
strong s-framed languages, like the Germanic languages, where v and Path are not
required to form one word, and, thus, allow constructions like complex adjectival
resultative constructions. Second, there are weak s-framed languages, like Latin,
where v and Path must form one word and may disallow, hence, constructions like
adjectival resultative constructions. This distinction is accounted for in terms of a
Path-to-v (PF) Raising operation for weak s-framed languages, which creates a
complex head. A three-way, gradual typology emerges encompassing strong s-
framed languages (no Path-to-v Raising required, no Path-v adjacency required),
weak s-framed languages (Path-to-v Raising required, no Path-v adjacency required),
and v-framed languages (Path-to-v Raising required, Path-v adjacency required).
. Methodology
.. The advantages of a theoretical approach to the grammar
of unspoken languages
This is, primarily, a study on theoretical linguistics, in particular, on how to handle
cross-linguistic variation in generative grammar. It is, secondarily, a study on Latin.
Since it has become a bit of a tradition in works like the present one to justify this
seemingly unnatural marriage, I shall also say a few words about it.
Needless to say, the main problem in doing generative grammar on an unspoken
language is the lack of native speakers. In particular, we do not have access to
competence, but only to performance, since we cannot elicit grammatical judge-
ments. Beyond the use of what ancient grammarians said about their language
(cf. Varro’s De lingua latina, On the Latin language) or any non-native competence,
built after years of exposition to the texts (Pfister , Miller :), we must rely
on closed corpora. But these data are, of course, natural, not experimental, and
deny us the precious gift of negative evidence, i.e. the starred sentence. Moreover,
we cannot be a hundred per cent sure that what has survived up to our times in the
manuscripts is undoubtedly positive evidence and we can only confide in the
expertise of the philologists to provide us with reliable editions.
I would like to assuage the dramatic scenario just depicted by pointing out how
generative grammar, or any well-articulated theory, for that matter, can shed light on
Methodology
the grammar of ancient languages. Interestingly, É. Kiss () notes that there have
been two major approaches to grammatical descriptions of unspoken languages. The
traditional approach is inductive, in that it builds a description from the data
available in the closed corpus. More recently, theoretical approaches, which are
deductive in nature, formulate hypotheses couched within a general theory of gram-
mar, and validate them against the data of the corpus. While the inductive approach
has proved useful in ‘listing and interpreting the morphemes of a language’ (É. Kiss
:) and in making generalizations concerning the different levels of grammar,
such an approach is, by necessity, considerably less heuristic than a deductive
approach. Specifically, it is only when equipped with a theory that we are in a
position to look for particular constructions—since we predict that they are possible
or not—and that we can thus ask ourselves why a particular construction is not
attested in the corpus. In this way, a deductive approach compensates for the lack of
negative evidence characteristic of corpora.
This work provides a perspicuous illustration of the advantage of a deductive
approach in addressing data from unspoken languages. As an example, I will show in
Chapter that Latin does not feature complex adjectival resultative constructions, i.e.
constructions like Sue hammered the metal flat, in which flat encodes the final state
attained by the metal and hammered encodes the way in which Sue brings the metal
to that state. As far as I know, this claim about how argument structure is expressed
in this language has never before been made in the Latin linguistics tradition or
elsewhere. Importantly, although the claim is empirical and arrived at through a
thorough corpus search that I shall describe in section .., I would never have made
it were it not for the fact that, from a particular theoretical perspective presented in
Chapters and , complex adjectival resultative constructions are expected to be
allowed in languages like Latin (s-framed languages). The theory leads us to the data.
In turn, the empirical finding in Latin leads me to non-trivial empirical and theor-
etical questions: do other s-framed languages disallow these constructions? Is
Talmy’s (, ) typology to be refined? Can I accomplish the refinement
through the theoretical tools that I assume?
Iacobini (:) call the ‘relative homogeneity in the control of the written norm’
(my translation), applicable to Classical Latin in the broad sense, and, hence,
encompassing Silver Latin. In particular, these authors point out that ‘in the literary
texts of the first two centuries of the Empire the prevalent norm is that of the Golden
Age [i.e. the Classical period in the narrow sense].’ (Crocco Galèas and Iacobini
:; my translation). A second reason is my suspicion that Late Latin (from the
third century to the sixth century AD) shows important differences as far as the
empirical domain of this work is concerned, i.e. argument structure and, secondarily,
Aktionsart. It will become clear in Chapters and that Latin makes use of verbal
prefixes in expressing argument structure changes and that there is a non-trivial
relation between prefixation and telicity. However, as Haverling (:) con-
cludes in a monumental work on the Aktionsart properties of unprefixed vs prefixed
sco-suffixed verbs, the event-structural function of the prefix is clearly lost by the end
of the second century AD—see also Barbelenet :– for an early observation in
the same vein. Taking into account these two reasons, I adopt the working hypothesis
that the periods of Early and Classical Latin (in the broad sense) constitute a
homogeneous language stage as far as the morphosyntactic expression of argument
structure and Aktionsart is concerned.
Unless otherwise stated, the data have been extracted from the CD-ROM corpus of
the second edition of the Bibliotheca Teubneriana Latina (BTL, Tombeur ), a
digitalized collection of the highly reliable Teubner’s Latin text editions. I restrict the
corpus searches to the Antiquitas subcorpus of the BTL, since this subcorpus
corresponds to the relevant period described above, from the first written texts
down to the end of the second century AD. The Antiquitas subcorpus contains a
total of , sentences. The procedure I have used to retrieve the data from the
BTL consists in searches for combinations of particular elements within the men-
tioned Antiquitas subcorpus. For instance, a search for telic instances of the prefixed
verb advolo ‘fly onto’ could involve the search of the combination of the sequence
‘advol*’, which yields all the registered forms of the verb without the inflectional
endings, and telicity-signalling expressions such as subito ‘suddenly’.
Besides the BTL, I draw on the data and descriptions thereof found in the rich
tradition of works on Latin linguistics, from the nineteenth century onwards. Of
particular importance, also, are the Latin dictionaries: Gaffiot’s () Dictionnaire
Latin-Français and Lewis and Short’s () Latin Dictionary, available online at the
Perseus Digital Library Project (Tufts University; Crane ). I have also found data
in other online corpora, although I have always ascertained that the data were also
registered in the Antiquitas subcorpus of the BTL, and, accordingly, I have always
labelled them with the reference provided in the BTL. In particular, I have made use
of the Greek and Roman Materials database at the Perseus Digital Library Project, the
LacusCurtius database (University of Chicago; Thayer ) and the Itinera Elec-
tronica database (Université catholique de Louvain; Meurant ). For some of the
Structure
texts found in these corpora there is a translation available, which I have often taken
into account; however, I always provide a translation of my own for all Latin data, if
not otherwise stated.
. Structure
In Chapter I put forward a theory of argument structure and the syntax-
morphology interface. The theory to be presented pertains to the class of so-called
neo-constructionist theories, that is, theories where argument structure properties do
not emerge from lexical items, but are properties of the syntactic configurations built
by the computational system. I discuss three such previous theories that have
inspired my own, as well as Hale and Keyser’s (, , ) programme,
the first attempt (after Generative Semantics) to provide a syntactic explanation of
lexical facts.
In Chapter I deal with the syntax-morphology interface, adopting Embick and
Noyer’s (, ) and Embick’s () theories. I further adopt the idea that some
functional nodes may fail to be interpreted at PF, if the conditions for the insertion of
their exponents are not met, yielding a crashed derivation. Since those conditions are
stated as part of idiosyncratic, language-specific properties of the Vocabulary Items
of the nodes in question, the possibility of a natural explanation for cross-linguistic
variation emerges, based on the morphological properties of functional items.
Chapter attempts to show that Latin is an s-framed language, in the sense of
Talmy (): in predicates expressing a transition, the element conveying the
transition and the verb correspond to different phonological units. First I introduce
Talmy’s (, ) theory of transition events, and his distinction between
v-framed languages (like Romance, where the transition cannot be expressed as an
element different from the verb) and s-framed languages (like Latin). I make a
syntactic interpretation of Talmy’s theory and propose that the s-/v-framed distinc-
tion is to be accounted for in morphological terms: in v-framed languages Path has to
be linearly adjacent to v for Vocabulary Insertion to proceed. This makes it impos-
sible for v to associate with an independent root, giving rise to the effect that Path and
v form a portmanteau morph. In s-framed languages, on the other hand, Path does
not have to be strictly adjacent to v, which produces the effect that they are realized
through different exponents. After this theoretical introduction, I carry out an
investigation of the expression of events of change in Latin, and I show that this
language is indeed an s-framed one. I introduce data that, as far as I know, have not
been tackled before in the Latin linguistic tradition—e.g. Ground Unselected Object
Constructions or Pseudoreversatives. All the constructions receive a uniform analysis
based on the status of Latin as an s-framed language. Finally, within a scenario that
goes beyond Latin, I propose new hypotheses on the nature of phenomena like the
Locative Alternation.
Introduction
A neo-constructionist perspective
on argument structure
In this chapter I present the view of the lexicon-syntax interface that will be defended
throughout the book. I adopt a perspective often referred to as neo-constructionist
(Levin and Rappaport Hovav : ), where the computational system of the
language faculty creates structures independently of the semantic encyclopaedic
features of lexical items, and where the compositional semantics of those expressions
is directly read off the syntactic structure. The role of lexical items in the interpret-
ation of linguistic expressions is reduced to that of contributing their encyclopaedic
content. In section . I describe the two main types of theories of the lexicon-syntax
interface: the projectionist and the constructionist theories. In section . I examine
Hale and Keyser’s () theory of argument structure as a predecessor of three neo-
constructionist frameworks: the theory of relational syntax and semantics of argu-
ment structure put forward by Mateu (), the exo-skeletal model of event structure
by Borer (, b), and the Distributed Morphology model (Halle and Marantz
, Marantz , among others). In section . I put forward a model drawing on
the three models presented in section .. The central idea on which the theory is
built is the difference between elements conveying encyclopaedic content, roots, and
elements conveying grammatical content, functional heads.
syntactic environment in which it appears. The first point of view is at the origin of
theories aiming at providing a necessary and sufficient characterization of the
semantic elements involved in a given lexical item that are relevant when determin-
ing its syntactic environment. Such theories are particularly concerned with the
design of appropriate lexical semantic representations that adequately register
those semantic elements crucial in determining the lexical item’s syntactic proper-
ties.1 To put it in Rappaport Hovav and Levin’s (: ) terms, ‘on this approach,
the lexical property of a verb that is taken to determine its syntactic behaviour is its
meaning (e.g., Levin ; Levin and Rappaport Hovav ; Pinker )’. Com-
plementarily, if the theory does not endow lexical items with a formal apparatus
marking the syntactic expression of their semantic information, it will design the
algorithms necessary for deriving the lexical item’s syntactic environment from the
mentioned syntactically relevant semantic elements. And, of course, it is possible that
both a representation of grammatically relevant properties of the lexical item and a
lexicon-(morpho)syntax mapping algorithm are provided.
Conversely, there are theories of the lexicon-syntax interface that try to uncover
which syntactic structures give rise to what semantic interpretations within a given
syntactic domain, taking in that way some of the weight of the semantic interpret-
ation from the lexical item itself and carrying it over to the syntax—in other words, to
functional categories and functional structure. In such theories, there is no need for
rich lexical semantic representations accounting for the lexical item’s syntactic
behaviour, or special algorithms relating the relevant aspects of meaning to morpho-
syntactic expression. There exists, however, a requirement of accurately describing—
often after enriching—the functional architecture of a sentence so as to account for
its syntax and its compositional semantics, abstracting from the conceptual content
of the lexical items it embeds. Borer (: ) calls the theories of the former kind
endo-skeletal theories, and those of the latter, exo-skeletal theories. This is not, of
course, the traditional nomenclature. Thus, Levin and Rappaport Hovav () call
the former theories projectionist, because the structure is projected from the lexical
item, while the latter are constructionist, because the compositional semantic and
syntactic properties are part of the construction, and not of the lexical item embed-
ded within. Borer’s () terms are based on the two basic types of skeletons we find
in the animal kingdom: the endoskeleton or internal skeleton, found in vertebrates,
and the exoskeleton or external skeleton, found typically in arthropods. Similarly, in
endo-skeletal theories, the structure is considered to be built from the inside, that is,
1
In fact, it was within this kind of theory that the difference between grammatically relevant and
grammatically irrelevant semantic aspects of a lexical item was first pointed out (see Pinker ;
Rappaport Hovav and Levin ). However, in some cases some lexical semantic aspect has been
considered grammatically relevant by one researcher and grammatically irrelevant by another one. See,
for instance, Mateu’s (: ff.) refutation of Levin and Rappaport Hovav’s () semantic notion of
internal/external causation as determining the realization of arguments.
Endo-skeletal vs exo-skeletal approaches
from the lexical items embedded in the structure, as it is through the properties of
these lexical items that linguistic expressions are built. In this sense, lexical items and
their properties constitute the structure’s skeleton (an endoskeleton). On the other
hand, in exo-skeletal theories functional structure is the skeleton—an exoskeleton—
of linguistic expressions, in that it is this structure that determines the (compos-
itional) semantic and syntactic features of the sentence. In turn, lexical items are
embedded within this exoskeleton. Here I will adhere to Borer’s terminology, and
I will reserve the term constructionist for the exo-skeletal models where the syntactic
structure corresponds, almost entirely, to lexically stored constructions. The exo-
skeletal models where structure is built by the computational system, that is, where
constructions are not primitive entities, will be called neo-constructionist—generative-
constructivist in Ramchand’s (: ff.) terms. Importantly, the discussion in the
present section is based almost entirely on the contrast between endo-skeletal and
neo-constructionist approaches, in spite of the use of the term exo-skeletal in
referring to the latter.2
In order to get a taste of how these general perspectives work out the relationship
between lexical semantics and syntax, let us have a look at the way they would
approach that relationship in the following sentence:
() The elephant broke the mirror.
In considering the relationship between the meaning of break and the syntactic
properties of the sentence it appears in, an endo-skeletal approach postulates a lexical
unit (stored among many others in some kind of lexicon), break, provided with a set
of idiosyncratic formal properties: a category V, a lexical semantic representation
and, perhaps, a subcategorization frame. The lexical semantic representation could
assume a variety of formats, for instance some kind of list of the theta-roles of the
participants in the event described by break. In the case of break two theta-roles
would be listed: the Agent or breaker and the Patient or thing broken. If a subcat-
egorization frame were also provided, it would contain information about the
insertion context of break, such as þ__NP, encoding the obligatoriness of an NP
in object position when break is inserted (all verbs have an (overt) subject in English,
so there would not be a need to state that for break).3 In most endo-skeletal models,
2
Examples of endo-skeletal theories are Williams (); Kaplan and Bresnan (); Pesetsky ();
Di Sciullo and Williams (); Wunderlich (); Grimshaw (); Van Valin (); Levin and
Rappaport Hovav (); Rappaport Hovav and Levin (); and Reinhart (, ), among others.
Examples of exo-skeletal theories are Ghomeshi and Massam (); Arad (, , , );
Kratzer (); van Hout (, ); Ritter and Rosen (); Ramchand (, , ); Travis
(); McIntyre (); Åfarli (); Starke (); Lohndal (); and De Belder and van Craenen-
broeck () among others.
3
Of course, break may appear in an intransitive context where the subject is the thing being broken, as
in The mirror broke. The endo-skeletal approach would probably posit mapping mechanisms (lexical or
not) to derive one alternant from the other.
A neo-constructionist perspective on argument structure
however, some general mapping mechanisms, either lexical or syntactic, convert the
list [Agent, Patient] into syntactic knowledge—both its phrasal category and its
syntactic function:
() Mapping from thematic roles to grammatical functions
a. Agent ! An NP subject (The elephant)
b. Patient ! an NP object (the mirror)
Such an approach predicts that, like break, other verbs with the same theta-grid
would resolve the mapping in the same way; crush, for instance, would incorporate
the same solution, at least as far as the sentence in () is concerned: [crush], [‘destroy
by exerting a hard pressure’], [Agent, Patient]. This is indeed the case: The elephant
crushed the mirror.
An exo-skeletal approach conceives of the structural properties of the expression
in () as responsible for some aspects of its semantic interpretation, such as the
notion of Agent or Theme, or its aspectual properties, and of its syntax, such as the
presence of an object or of a subject. Many of these structural properties are covert, of
course. In this approach, particular attention is paid to the presence of the same unit,
break, in other very different syntactic contexts, as in (), where the intended
meaning is ‘the elephant went in violently’:
() The elephant broke (*the mirror) in.
In contradiction of the prediction of the endo-skeletal approach, there seems to be no
possible projection of the Theme argument in (), an alleged idiosyncratic property
of break. The exo-skeletal approach would interpret the structure of the sentence in
() as disabling the appearance of the object, and would try to give an account of that
disallowance in terms of the syntactic structure. Probably, in the face of the avail-
ability of break in () and () the endo-skeletal account would propose two breaks, an
object-projecting break and a second lexical item break in, which would not count a
Theme within its theta-grid (hence, not projecting it in the syntax). The problem here
would be the failure to capture the generality that other verbs that, like break,
obligatorily project an object in certain structures (The elephant broke *(the mirror))
cannot project it when appearing with some particles. This is the case of smash,
another verb that cannot drop its object: The elephant smashed *(the mirror). Smash
is obligatorily intransitive when combined with through, as in () below, in the
interpretation in which the elephant is entering somewhere after going through
some entity (the sentence accepts the direct object in the interpretation in which
the elephant does not go through the mirror):
() The elephant smashed (*the mirror) through.
However, not only does the break case extend intra-linguistically, to other verbs
within the same languages, but also cross-linguistically. Thus, the break/break in
Endo-skeletal vs exo-skeletal approaches
alternation parallels the one found in Latin between rumpo ‘break’ and prefixed in-
rumpo “in-break” ‘break in’:
() Latin
a. Elephans *(speculum) rupit.
elephant.NOM mirror.ACC broke
‘The elephant broke the mirror.’
b. Elephans (*speculum) in-rupit.
elephant.NOM mirror.ACC in-broke
‘The elephant broke (*the mirror) in.’
If, as is probably assumed within the endo-skeletal approach, there are two lexically
listed (although related, as I said before) breaks, accounting for their different
argument structure properties, the question is why a similar listing obtains in a
different lexicon, namely that of Latin.4 Conversely, the exo-skeletal approach would
develop a theory of sentential architecture apt to host a position for the object in the
case of (), without resorting to any idiosyncratic properties of break. In doing this, it
might run the risk of either creating nonexistent structure or overgenerating.
Within such a scenario, a fundamental asymmetry arises between the articulations
of these two types of theory. While in the former type, the endo-skeletal, the interface
between the lexicon and the syntax is non-trivial, in the sense that it is the semantic
properties of lexical items that derives their syntactic properties, in the exo-skeletal
type the interface is considerably reduced, if it exists at all. In this sense, Borer (b:
) points out that ‘[c]ontrary to common assumptions, there is, in actuality, no direct
interface, as such, between the conceptual system and the grammar, in that proper-
ties of concepts do not feed directly into the determination of any grammatical
properties’. In attributing all non-purely conceptual semantic aspects of linguistics
expressions to the syntactic structure, paradoxically, exo-skeletal theories turn out
not to be theories of the lexicon-syntax interface any more, as they do not envision
any such interface. They attempt to explain problems of the relationship between
lexical semantics and syntax, dividing what has traditionally been packed together
as lexical semantics into compositional semantics and conceptual semantics, and
rethinking the former as an emergence of syntactic structure. Thus, in developing the
appropriate functional architecture, which is often phonologically covert, they seek to
explain the syntactic and compositional-semantic properties of the sentence.
The crucial difference just exposed is directly related to a difference in how each
type of theory conceives of the minimal units the syntax plays with. As exemplified
4
Within a classical constructionist approach (Goldberg , ), where constructions are primitive
lexically listed units, the cross-linguistic facts are difficult to accommodate. In general, lexical marking is a
problem when cross-linguistic parallels are found, since they remain, within frameworks that resort to
lexical marking, as mere coincidences.
A neo-constructionist perspective on argument structure
above, endo-skeletal theories typically work with units that, besides incorporating the
Saussurean relationship between the phonological information and the conceptual
information, also make explicit the semantic components (theta-roles, event struc-
ture, aspectual features, etc.) that are taken to be relevant for the construction of the
syntactic environment in which the lexical item appears. These theories must also
provide some formal code determining the syntactic behaviour of the lexical item,
which is either predictable from the grammatically relevant aspects of meaning or
not. By contrast, although there might be differences among various models, in exo-
skeletal theories lexical items are typically units endowed exclusively with encyclo-
paedic content, given that grammatically relevant aspects of meaning are claimed to
emerge from structural properties of the sentence.5
We use the term argument structure to refer to the syntactic configuration projected by a
lexical item. It is the system of structural relations holding between heads (nuclei) and their
5
In this vein, Goldberg (: ) notes the need ‘to distinguish the semantics of argument structure
constructions from the verbs which instantiate them, and to allow the verbs to be associated with rich
frame-semantic meanings’. In Goldberg’s () constructionist framework, in fact, the semantics of verbs
are reduced to Frame Semantics (Fillmore , ).
Three neo-constructionist theories and a predecessor
arguments within the syntactic structures projected by nuclear items. While a lexical entry
is more than this, of course, argument structure in the sense intended here is nothing more
than this. Hale and Keyser (: )
The syntactic character of their theory is based on the idea that the same principles
that operate in syntax, accounting for both grammatical and ungrammatical syntac-
tic patterns, can also explain patterns in the lexicon, such as lexical gaps, argument
structure alternations, or the syntactic behaviour of verbal classes. In particular, Hale
and Keyser () propose that argument structure types reduce to four basic
syntactic configurations defined by the projecting properties of their lexical heads:
() Hale and Keyser (: )
a. [h h cmp] (V)
b. [h spc [h h cmp]] (P)
c. [h* spc [h* h* h]] (A)
d. h (N)
The configuration in ()a is headed by a category, h, that only takes complements. In
()b the heading category takes both a specifier and a complement. In ()c, h takes
only a specifier and must thus combine with an ancillary category (h*) of the type of
()a to project it. Finally, the configuration of ()d corresponds to a category with
zero valency, not taking any arguments. In the unmarked case, the configurations
in () are realized, respectively, as V(erb), P (adposition), A(djective) and N(oun),
in English. In () there is an example of an argument structure configuration,
namely, that corresponding to the predicate clear the screen, headed by the dead-
jectival verb clear:
V VP
NP V’
N V AP
screen A
clear
The A lexical head clear projects a specifier (screen) thanks to the fact that it is taken
as complement by a V head, characterized by the selection of a complement (see ()a
and ()c). In turn, the whole VP is taken as complement by another higher V which
transitivizes the predicate. In order to account for the fact that the verb clear is
pronounced as such, Hale and Keyser propose that this verb is formed by an instance
A neo-constructionist perspective on argument structure
of head movement that they call conflation, which takes clear up into the intermedi-
ate V and finally into the highest V:
V VP
cleari V NP V’
N V AP
screen ti V A
ti
In this theory thematic roles are not primitive, but interpretations of the positions
occupied by arguments in the configurations (and see Hale and Keyser : ff.
for considerations on the semantic interpretation of their argument structure
configurations).
The following is an example of how independently postulated syntactic principles
account for patterns of lexical well-formedness: while it is possible to derive a
predicate such as clear the screen as depicted in the next example, it is impossible
to derive such predicates as *to metal flat, meaning ‘to flattened (the) metal’, or *to
spear straight, meaning ‘to straightened (the) spear’ (Hale and Keyser : ). This
is due to the fact that, in order to derive these predicates, conflation should apply
from specifier position into the upper head:
V VP
metali V NP V’
N V AP
ti A
flat
The ECP, a syntactic principle, can then explain a lexical fact: the non-existence, in
English, of verbs whose root designates an object submitted to a change of state and
which co-appear with an adjective expressing the resulting state. Crucially, this
explanation depends on the assumption that there is a level of representation of
the verb where its argument structure is syntactically displayed.
The scenario depicted seems to fit the characterization of a neo-constructionist
system, since argument structure properties and interpretation of arguments hang
on syntactic projections. However, two features of the theory militate against this
qualification. First, the status of the category A(djective) and, second, the l-syntax/-
s-syntax distinction. Since Mateu () accomplishes a successful theoretical
solution for the first problem, I leave the consideration thereof for section ..
and I concentrate now on the second problem. Consider first the following excerpts,
which form part of definitions of argument structure included in different works
by Hale and Keyser and which explicitly assume a lexical encoding of syntactic
properties:
() Hale and Keyser (: )
‘[A]s a matter of strictly lexical representation, each lexical head projects its
category to a phrasal level [ . . . ].’
6
See also Hale and Keyser (: ), where their research project is defined as stemming ‘from a
general program of study implied by the Projection Principle (Chomsky ) and the notion that syntax is
projected from the lexicon’. See also Hale and Keyser (b: footnote ) where they state that verbs must
be listed in the lexicon, much as their formation is syntactic. This is how they explain why not all
imaginable unergative birthing verbs are possible: The mare foaled, The shad roed, ?The kangaroo joeyed,
*The cat kittened, *The sow pigleted.
Three neo-constructionist theories and a predecessor
This is the fate of roots. What about the rest of the components of the argument
structure configuration? Hale and Keyser () propose that the argument nodes,
such as the one marked with DP in (), are variables where fully fledged phrases are
inserted at d-structure. The rest of the nodes are eliminated by some node-pruning
operation. Both the node-pruning mechanism and the fact that argumental positions
are refilled with DPs at d-structure clearly argue for the existence of some break
between l- and s-syntax. If, in addition to this, we take into account the fact, observed
by Hale and Keyser (: ), that there is no evidence that conflating elements
leave traces, in the s-syntactic sense of the term, we get quite a separate cycle of
syntactic computation.7 Besides the fact that l-syntax and s-syntax are different
because they constitute different cycles and l-syntax includes at least one
operation—conflation—that is not attested in s-syntax, Hale and Keyser resort to
an ontological difference between both based on the lexical, i.e., stored or static,
character of the former and the dynamic character of the latter, as can be gathered
from the following quote:8
The idea that the grammatical properties of a lexical item are syntactic in character, and that
they include dependencies of the type represented by the trace-antecedent relation, should not
be taken to imply that the use of a lexical item entails the actual application of movement rules
in processing or producing the sentence. Thus, the use of the verb saddle does not involve
performing a derivation, relating () and () [two representations of the verb saddle]. Rather,
the representation embodied in () and () is a static lexical representation of the relevant
grammatical properties of the verb saddle. It is, by hypothesis, present in the linguistic
knowledge of speakers of English who happen to know the verb. But it is not ‘accessed’ at
s-syntax. It is not visible there. Hale and Keyser (: )
I point out that the alleged staticness or ‘storedness’ of l-syntax is in prima facie
contradiction with its syntactic character. In particular, if knowledge of the syntactic
behaviour of a lexical item is really syntactic, then it cannot be in any case different
from the knowledge involved in the derivation of a sentence, as it should itself involve
a derivation. On the other hand, the knowledge involved in the derivation of a
sentence should be as static as Hale and Keyser claim l-syntax to be, if it is seriously
7
And observe that, before their revision of their concept of conflation in the third chapter of Hale and
Keyser (), the original sites of conflating elements could be occupied by overt material in s-syntax, as
in the account of cognate objects (like dance in She danced a silly dance). This insertion would add to the
counter-cyclicity of l-syntax with respect to s-syntax. See Haugen () for discussion.
8
Travis (: ), for instance, after accepting the Halekeyserian computational analysis of denom-
inal verbs like shelve (see section ..) states the following: ‘My conclusion will be that there is a principled
distinction which is not surprising—one [an l-syntactic process] appears to happen in the lexicon and is
therefore idiosyncratic, while the other [an s-syntactic process] arguably happens in the computational
system (i.e. syntax) and is therefore productive.’ The relevant point here is the opposition of ‘lexicon’ vs
‘computational system’.
A neo-constructionist perspective on argument structure
assumed that both l- and s-syntax pertain to the domain of competence and not to
that of performance. Similar remarks could be made of the following, later excerpt:
We will continue to use these diagrams, where convenient, with the understanding that they
are abstract informal representations of argument structure properties and not the represen-
tation of any actual point, initial, medial, or final, in the derivation of a verbal projection—they
could not be that, under the assumptions of a ‘bare phrase structure’ theory of lexical
and syntactic projection (Chomsky ) or under the assumption of ‘late insertion’.
Hale and Keyser (: )
Here they point out the assumptions of Bare Phrase Structure (Chomsky ) and
Late Insertion (Halle and Marantz ), as these refer typically to properties of
s-syntactic derivations (to be precise, Late Insertion refers to derivations in the
phonological branch of the derivation), and argument structure configurations do
not comply with them. They still oppose ‘abstract’ as a property of l-syntactic
representations and ‘actual’ as a property of s-syntactic derivations, and the same
fallacy obtains.
This statement is the natural effect of the conceptual necessity that those aspects of
meaning that are compositional must be so in syntactic terms, while those aspects
of meaning that are not compositional cannot be stated in syntactic terms. That is,
semantic construal cannot be at the same time syntactically non-transparent, and
conceptual content cannot be at the same time syntactically transparent. In this way,
there is a strong (and natural) correlation between computation (syntax) and com-
positional meaning, on the one hand, and the non-computational elements of
linguistic expressions and non-compositional meaning, on the other. In conformity
with this statement, Mateu makes a crucial distinction between relational and non-
relational elements. Relational elements form a closed set, and constitute the articu-
lators of argument structure configurations, in that, besides being endowed with
certain highly abstract semantic content, they interrelate the building blocks of the
structure. Non-relational elements crucially do not have syntactic properties (not
even syntactic category), only conceptual ones: they cannot project a specifier or a
complement.
Three neo-constructionist theories and a predecessor
The relational heads proposed in Mateu () are basically two, although the
second comes in two varieties: one head, [r], is semantically interpreted as a non-
eventive relation, and projects both a complement and a specifier; the other is an
eventive head projecting a complement but only optionally projecting an external
argument (EA) as the specifier of some higher functional head (F). The EA-
projecting eventive head is [R], the source relation, while the one that does not
project it is [T], the transitional relation. These three heads are specified for a
value. If we leave this non-configurational property aside, the interpretation of [R],
[T] and [r] can be said to emerge purely from configuration. In particular, these
heads are to be found in the following configurations (F = functional head introdu-
cing the EA; X = a non-relational element):
As can be gathered from () to (), the þ value for [R] is associated with agentivity
(e.g., in John rolled deliberately vs John stank), the þ value for [T] is associated with
dynamicity (e.g., in The ball rolled vs John lived), and the þ value for [r] is related to
change and telicity (e.g., in John killed the horse or John died vs John pushed the horse
or The ball rolled). The combinations of () to () are not all the logical ones given
the number of relational heads and the number of values. As observed by Real
Puigdollers (: ), there are two surprisingly similar gaps in the paradigm of
transitives and in the paradigm of unaccusatives:
() *[F X . . . F . . . [-R [X [þr X]]]]
A neo-constructionist perspective on argument structure
Verbs incorporating an abstract preposition are, in Hale and Keyser’s () theory,
different from those incorporating an adjectival head, like clear. The difference is
related to the fact that only the latter are claimed to enter in the so-called Causative
Alternation:
() Hale and Keyser (: and )
a. The screen cleared.
b. *The book shelved.
c. *The horse saddled.
Configurationally, transitive clear has two V layers, a transitivizing one and an
unaccusative one. Thus, if the outer layer is taken off, the structure is still a verb,
and its specifier counts as the surface unaccusative subject (see ()a and ()a). The
presence of the internal V layer is due to the fact that the head A, which projects only
a specifier, needs the complement-projecting head V to project that specifier. On the
other hand, verbs involving a P projection have only one V layer, which is both the
verbalizing head and the transitivizing head (P, in projecting both a complement and
a specifier, does not need any other head to project) (see ()b, ()c and ()b):
() Hale and Keyser (: and )
a. [V the screen [V V A (= clear)]]
b. [V V [P the books/the horse [P P [N shelf/saddle]]]]
Mateu (), however, following Kiparsky (), argues that the facts in () are
due not to a grammatically encoded distinction, but to world knowledge. Thus, if the
action described by the predicate can be understood as non-agentive, locatum/
location verbs may license an unaccusative use (see (), where the helicopter is a
self-propelled object); the same applies to deadjectival verbs like clear, which may
(see ()d) or may not (see ()b) appear in unaccusative predicates on the grounds
of the same non-agentive/agentive reading:9
() Catalan; Mateu (: )
L’helicòpter aterrà tard.
the=helicopter (to)landed late
‘The helicopter landed late.’
9
In Acedo-Matellán (a) I provide more examples of uncontroversially locatum/location verbs
which, depending on the interpretation, may or may not enter into the Causative Alternation. Thus, for
instance, Cat. locatum em-perlar “in-pearl”, incorporating the prepositional prefix en- ‘in’, may be used to
mean ‘bead (a necklace)’ or ‘cover with pearl-like elements, like dew drops’. Thus, in the former use
emperlar invokes an agent-controlled scene, but not in the latter. Accordingly, emperlar may only appear as
intransitive in the latter use.
A neo-constructionist perspective on argument structure
() Levin and Rappaport Hovav (: –), in Mateu (: )
a. The waiter cleared the table.
b. *The table cleared.
c. The wind cleared the sky.
d. The sky cleared.
Once these facts have been acknowledged, there is no evidence that locatum/location
and deadjectival verbs differ grammatically. More generally, there remains no evi-
dence for a distinction between structures ()b and ()c. In particular, the h head
in ()c, which is defined as the head projecting a specifier but no complement, and
which is unmarkedly realized as A in English and many other languages, is non-basic.
Instead, it is amenable to a decomposition into an [r] relation (P in Hale and Keyser’s
terms) and a non-relational element (N in Hale and Keyser’s terms). I mention again,
lastly, what I pointed out in section ..: that this move has a welcome consequence
not sufficiently emphasized by Mateu (). Specifically, Mateu () eliminates
the undesirable situation of having an element (h in ()c) be relational and convey
conceptual content, simultaneously. In that sense, Mateu’s () theory can be
argued to fulfil the neo-constructionist desideratum of neatly separating roots
(non-relational elements) from the material able to create structure (relational
elements). See also Amritavalli and Jayaseelan (), Amritavalli (), and
Kayne () for the proposal that adjectives are to be analysed as non-primitive
categories, involving the combination of a non-relational element and an adposi-
tional element.
Finally, some lexicalist traces can be found in Mateu’s () theory that I wish to
refute. Turning back to the discussion on the telic nature of location/locatum verbs
(see () above), he points out some apparent counterexamples:
These examples would jeopardize his proposal that both location and locatum verbs
incorporate a [þr] relation, inducing telicity. With respect to examples like ()b,
Mateu observes that the licensing of the durative adverbial is due to a measurement
of the resulting state: in this case, durant un minut ‘for a minute’ expresses the time
span spent by the bird in the cage after having been caged therein. With respect to
examples like ()a, Mateu points out that the non-relational element involved
refers to a mass entity, in this case flour (farina), and that this fact licenses an atelic
Three neo-constructionist theories and a predecessor
reading of the predicate. Thus, since the root does not refer to a bounded entity, the
action of putting that entity somewhere (the meatballs) cannot be measured out:
enfarinar “in-flour” ‘flour’ would turn out to be like ruixar ‘spray’, which can also
license an atelic reading for exactly the same reason in John sprayed the wall with
paint for five minutes (Mateu : ). Crucially, though, enfarinar cannot be said
to involve a [-r] relation—present in verbs like empènyer ‘push’—which would, on
the other hand, account for its atelic reading straightforwardly. The enfarinar/
empènyer ‘flour’/‘push’ dissociation and the enfarinar/ruixar ‘flour’/‘spray’ associ-
ation are based on diagnostics such as the following, involving licensing of adjectival
passives:
() Catalan; Mateu (: –)
a. Les mandonguilles estan enfarinades.
the meatballs PFV.be.PL (in)floured
‘The meatballs are floured.’
b. La paret està ruixada de pintura.
the wall PFV.be.PL sprayed of paint
‘The wall is sprayed with paint.’
c. *El carro està empès.
the cart PFV.be.PL pushed
According to this test, verbs like enfarinar pattern with verbs like ruixar in involving
a final state and licensing thereby the adjectival passive construction; on the other
hand, verbs like push, which do not involve a final state, disallow the adjectival
passive construction. Note, however, that the discussion is set, literally, in terms of
verbs, that is, lexical units, and in terms of what they involve as such. My claim here is
that neither does enfarinar ‘flour’ necessarily involve a [þr] head nor does empènyer
‘push’ necessarily involve a [-r] head. Accordingly, assuming Mateu’s primitives,
enfarinar ‘flour’ can be claimed to reflect either a [þR [X [þr X]]] configuration, in
which case a change of state is readily interpreted and telicity can thereby emerge, or
a [þR [X [-r X]]]] configuration, in which case no final state is entailed to be attained
and atelicity arises. I believe that what the diagnostics in () is really showing us is
that a very special context is needed for empènyer ‘push’ to be interpreted as
telic/change of state, unlike enfarinar ‘flour’ and ruixar ‘spray’. Thus, while it is
possible to conceive of a (bounded) quantity of flour or spray which would qualify as
standard in defining an end state for a flouring or spraying event, respectively, it is
considerably more difficult to evoke a standard ‘pushedness’. However, it is not
impossible, as the next example from Kratzer (), shows:
() German; Kratzer (: )
Dieser Kinderwagen ist schon geschoben.
this baby carriage is already pushed.
A neo-constructionist perspective on argument structure
In Kratzer’s (: ) words, ‘[a] natural setting for [()] would be a factory that
produces baby carriages and employs workers whose job it is to push new baby
carriages a few times to test their wheels’.
More generally, I think that diagnostic tests like the one in (), involving the
licensing of particular constructions, are not diagnostics about the membership of a
certain verb in a particular grammatically defined class: they could not be, once an
exo-skeletal perspective has been adopted, where category-free roots are freely
inserted in the structures generated by syntax, and hence, the only reason a root
does not fit into a structure is an incompatibility between the semantics emerging
from the structure and the conceptual content of the root. The adjectival passive
construction illustrated in () most probably involves some grammatical formative
like Mateu’s () [þr] relation, but enfarinar ‘flour’ or ruixar ‘spray’ or, more
specifically, the roots involved in them, do not.
the flower
<e>E TP
the flower
wilt<pst><e>T AspQP
the flower
<e># VP
wilt
The same DP the flower moves to the specifier of EP, through that of TP, to provide
range to the eventive open value, <e>E. The open value for tense, <e>T, is assigned
range directly by the abstract past tense head feature <pst>, which triggers head
movement of the listeme wilt. The DPs assigning range to the relevant open values
receive an interpretation ‘as an entailment of the event structure’ (Borer b: ).
Thus, the specifier of AspQP is interpreted as Subject-of-quantity (in Tenny’s
terms, it measures out the event), since it is the subject of a quantity predicate, namely
AspQP. As a specifier of EP, the DP is interpreted as an Originator, as originating
the (wilting) event. These interpretations are, crucially, independent of the listeme
that ends up being the verb (wilt in ()). In this view, unaccusative predicates are
A neo-constructionist perspective on argument structure
10
The difference between unaccusative predicates and telic transitive predicates lies in the fact that the
latter involve two argumental DPs: the object, merged as the specifier of the aspectual head and the subject,
merged first as specifier of TP and then as specifier of EP.
Three neo-constructionist theories and a predecessor
11
A critique based on Starke (). In Starke’s () nanosyntactic theory the nodes of the syntax are,
in fact, individual features, so there is no need for a pre-syntactic generative narrow lexicon.
Three neo-constructionist theories and a predecessor
12
Ultimately, the difference between Path and Place could be argued to be purely configurational, a
single category p being interpreted as Place and an ulterior one being interpreted as Path. In turn, the
categories v and p could themselves be conflated into one relational head, the distinction derived also from
configurational properties. See Boeckx () for related discussion.
13
See De Belder () and De Belder and van Craenenbroeck () for an interesting proposal in
which roots are not present during the syntactic derivation, their properties (including inability to project)
being derived from a particular theory of the operation Merge and a reformulation of the Subset Principle.
For other proposals doing away with the relational/non-relational dichotomy in syntax, see Real Puigdol-
lers (); Acedo-Matellán (); and Acedo-Matellán and Real-Puigdollers ().
14
Marantz (, ) has recently rejected the idea that roots may occupy argumental positions. See
Acedo-Matellán () for an exposition of Marantz’s () arguments that roots can only be modifiers,
and a refutation of this proposal on empirical and theoretical grounds.
The present framework
Sue Voice’
Voice vP
v dance
b. Sue did a dance.
VoiceP
Sue Voice’
Voice vP
v a dance
c. Sue pushed the car.
VoiceP
Sue Voice’
Voice vP
pP vP
p the car v push
v PlaceP
Dinosaurs Place’
Place exist
b. Sue is in Barcelona.
vP
v PlaceP
Sue Place’
Place Barcelona
Place in
A neo-constructionist perspective on argument structure
Path PlaceP
The sky Place’
Place clear
b. Sue went to Barcelona.
vP
v = went PathP
Sue Path’
Path = to PlaceP
Sue Place’
Place Barcelona
() Stative transitive event
a. Sue loves peaches.
VoiceP
Sue Voice’
Voice vP
v PlaceP
peaches Place’
Place love
b. Sue kept the car in the garage.
VoiceP
Sue Voice’
Voice vP
v = kept PlaceP
the car Place’
Place the garage
Place in
The present framework
Voice vP
v PathP
Path PlaceP
the sky Place’
Place clear
b. Sue shelved the books.
VoiceP
Sue Voice’
Voice vP
v PathP
Place shelf
c. Sue put the books onto the shelf.
VoiceP
Sue Voice’
Voice vP
v = put PathP
Place on
A neo-constructionist perspective on argument structure
Some remarks must be made about how these configurations relate to syntactic facts.
First, I follow Hale and Keyser’s () or Mateu’s () proposal that unergative
predicates (see ()a) are underlyingly transitive predicates. Specifically, within the
present proposal, unergative verbs like dance correspond to a vP where Compl-v is a
root, and not a DP/NP. The structure of unergative verbs as transitives is forced by
the properties of the system: it is not possible for a functional head to project a
specifier without projecting any complement, since the first DP/root merged with a
functional head must be its complement (and roots are independently ruled out as
specifiers, as I pointed out in ...). Hence, unergatives must be transitives (that is,
they must feature a complement—a root).15
In Hale and Keyser (: ff.) transitive activity verbs like kick are provided
with the following lexical-syntactic configuration (where F stands for a functional
category introducing the external argument), identical to that provided for stative
verbs like love—see Mateu () and Acedo-Matellán and Mateu () for an
analysis along similar lines:
() Sue kicked the door.
[FP Sue [F’ F [vP v [PP the door [P’ P KICK]]]]]
The P head in the above configuration is to be understood as involving a relation
of central coincidence. Hale and Keyser paraphrase predicates like kick someone as
‘give someone a kick’. It is hard to accept, though, that stative predicates like love
and dynamic, activity predicates like kick should involve the same syntactic
configuration, much as both types of predicates are atelic—a fact captured, within
this system, through the absence of the Path projection. A basic difference, for
instance, is that stative predicates can hardly drop their objects, while activity
predicates do so with relative ease. In fact, stative verbs seem to be forced to adopt
an activity interpretation when they are used intransitively, as in the second
example with love:
() Object drop in kick vs love
a. Sue kicked for several minutes.
b. ??Sue loved for years.
Following an idea in Marantz (), I consider transitive activity predicates such
as kick the door as involving an unergative configuration, the verbal root being
merged as the sister of v. The overt object is in fact an adjunct merged with the
vP through a null preposition interpreted as a central coincidence relation (see
also ()c). The adjunct-hood of this DP is what explains the fact that it can be
omitted:
15
The same rationale underlies the treatment of particles as ‘unergative’ prepositions. See section ....
The present framework
16
In the representations I have also abstracted away from other movements, for instance movement of
the internal argument for case-reasons (to the second Spec-Voice or to Spec-T).
A neo-constructionist perspective on argument structure
Similarly, within the domain of Latin prepositional prefixes, the separation of the
conceptual content of the prefix and the structure in which it is embedded finds an
echo in Carvalho’s (: ) distinction between signifié lexical and signifié cate-
goriel of the prefix.17
The dissociation of prepositions into a functional and a non-functional element
straightforwardly implements the well-established idea that particles are intransitive
prepositions (see Cappelle : ff. and references cited therein). In particular,
while PPs like on the shelf correspond to PlaceP structures in which the root of the
preposition is adjoined to Place and Compl-Place is a DP (the shelf), particles like on
correspond to PlacePs where the root of the preposition sits directly at Compl-Place.
The difference is illustrated below:
() The books (are) on the shelf.
[PlaceP The books [Place’ [Place Place ON] the shelf]]
() The lights (are) on.
[PlaceP The lights [Place’ Place ON]]
17
I conjecture that the fact that inventories of adpositions contain much fewer elements than those of
nouns is due to the fact that the set of possible or cognitively salient spatial relations, conveyed by
adpositions, is much smaller than the set of entities, conveyed by nouns. For more discussion on the
functional or lexical status of P, see Koopman () or Den Dikken (), among others. See also
Roßdeutscher () for an analysis of contentful prepositions as involving a root and a functional head p.
The present framework
Thus, particles, whether affixal (as in Latin) or not, turn out to be, specifically,
unergative prepositions, as illustrated in () (see Kayne ).
A root can also adjoin to v. Thus, the roots DANCE and HAMMER are adjuncts to v in
()a and ()b, respectively:
() Root-adjunction to v
a. Sue danced into the room.
vP
v PathP
v dance Sue Path’
Path = to PlaceP
Sue Place’
Place the room
Place in
b. Sue hammered the metal flat.
VoiceP
Sue Voice’
Voice vP
v PathP
Path PlaceP
the metal Place’
Place flat
Root-adjunction to v, which will be crucial for understanding the data dealt with in
this work, is designed to capture so-called lexical subordination constructions (Levin
and Rapoport ), that is, constructions involving a complex event where the main
event is identified with an accompanying co-event. Thus, for instance, in ()a the
unaccusative event whereby Sue enters the room is accompanied by a subordinate
event of dancing (although the dancing, it should be noted, is not linguistically
represented as a separate event, that is, through a separate v head). For similar
treatments of lexical subordination, see Embick (); McIntyre (); Zubizarreta
and Oh (); and Mateu (b, ), among others. For a reformulation in
terms of Bare Phrase Structure, see Acedo-Matellán ().
A neo-constructionist perspective on argument structure
c. Figure: a DP at Spec-Place
Dinosaurs existed, The sky cleared, Sue went to Barcelona, Sue is in
Barcelona, Sue kept the car in the garage, The strong winds cleared the
sky, Sue shelved the books, Sue put the books on the shelf
d. Central Ground: a DP or root at Compl-Place when no PathP is projected
Dinosaurs existed, Sue loves peaches, Sue is in Barcelona, Sue kept the car
in the garage
e. Terminal Ground: a DP or root at Compl-Place when PathP is projected
Sue went to Barcelona, The sky cleared (in five minutes), The strong winds
cleared the sky, Sue shelved the books, Sue put the books on the shelf
f. Measurer: a DP raised from Spec-Place to Spec-Path
Sue went to Barcelona, The sky cleared (in five minutes), The strong winds
cleared the sky, Sue shelved (the) books, Sue put (the) books on the shelf
g. Co-event/Conformation: a root adjoined to a functional category (v or Place)
Sue danced/tiptoed into the room, Sue hammered/trod the metal flat, The
book was in/on the box
These interpretations are in part localistic and in part aspectual, that is, Aktionsart-
related. The notions Figure and Central or Terminal Ground are localistic. The
Figure, in Talmy’s () terms, is the entity that is located or moving with respect
to some other entity, which is the Ground. For instance, Sue is a Figure and Barcelona
is a Ground both in Sue went to Barcelona and Sue is in Barcelona. The relation
between Figure and Ground can also be metaphorical, in terms of the predication of
some property: the Figure is an entity to which some property, encoded by the
Ground, is ascribed. Thus, the sky and clear are, respectively, a Figure and a Ground
in The sky cleared in five minutes and in The sky is clear.
The Ground, in turn, can be either a Central Ground or a Terminal Ground, a
localistic-aspectual distinction. A Central Ground corresponds to a location/state
that corresponds to a static description, as in The sky is clear. A Terminal Ground
corresponds to a final or resulting location/state. For instance, in Sue went to Barcelona
and The sky cleared in five minutes it is entailed that Sue ends up in Barcelona and that
the sky ends up in a pragmatically defined state of clearness after five minutes.
The Originator, the Effected Object, and the Measurer are event-structural
notions. An Originator is the entity that originates the event, as, for instance, is
The strong winds in The strong winds cleared the sky. An Effected Object is an entity
that comes into existence or disappears as the event evolves. For instance, in Sue
danced, the root DANCE, an Effected Object, refers to the activity of dancing, which
unfolds along with the event introduced by v. In Sue did a dance, the DP a dance
is the Effected Object, with the same interpretation. Lastly, a Measurer, a DP at
Spec-Path, is an entity that induces a measure for the transition introduced by
The present framework
PathP. Thus, for instance, in Sue shelved the books in five minutes or The sky cleared
in five minutes, the books and The sky are Measurers (they move to Spec-Path from
their original Spec-Place position, where they are interpreted as Figures) in that they
establish a measure for the events of shelving and clearing. Thus, these events will be
completed as soon as the entities denoted by the Measurers attain the location/state
denoted by PlaceP, that is, when all the books denoted by the books are shelved and
when the whole entity of the sky denoted by The sky is clear. However, note that I also
call Measurer a non-quantity DP like books in Sue put books on the shelf or Marine
life in Marine life swam into the cave for hours. In these predicates there is also a
transition encoded by PathP, but since the quantity conveyed by the object is not
definite, telicity cannot arise. See section ... for more details on the relation
between Path and (a)telicity and the interpretation and syntax of the Measurer.
As I pointed out in section ..., the roots adjoined to functional categories, like v or
Place, provide a conceptual specification of their abstract meaning. A root adjoined to v
is thus interpreted as a (Manner) Co-event in that it specifies the way in which the event
introduced by v is carried out. Thus, in Sue hammered the metal flat, the externally
originated event of change of state (of a metal which becomes flat) is identified with a
hammering activity, since v forms an adjunct structure with root HAMMER. A root
adjoined to Place identifies the type of spatial relation that a Figure holds with respect
to a Ground, its Conformation, in Talmy’s () terms (see section ...).
I point out, finally, a crucial difference between Mateu’s () theory and the present
theory, which concerns the interpretation of functional heads (relational heads in
Mateu’s terminology). Recall from section .. that relational heads are endowed
with either a þ or a - value, characterizing agentivity/non-agentivity (for R), transi-
tion/non-transition (for T), and telicity/atelicity (for r). Recall, also, that among struc-
tures featuring the r relation, two structures were missing in Mateu’s () model:
I claim that to the extent that the present account eliminates (non-configurational)
features in the interpretation of relational heads, the non-existence of the above
combinations is explained away. With respect to (), since I have not taken agentivity
to be linguistically represented I do not make a difference between þR (Sue sings: [F
Sue . . . F . . . [þR SING]]) and -R (Sue stinks: [F Sue . . . F . . . [-R STINK]]). Thus, I have
no non-existing combination to account for. As regards (), the þT/-T difference
relates to a dynamic/stative difference. However, I do not encode this difference on the
eventive head. Rather, an unaccusative predicate, if expressing a transition, is endowed
with a double p-projection (PathP) (cf. clear); if not expressing a transition, it is
endowed with a single p-projection (PlaceP) (cf. exist). In this scenario a configuration
equivalent to that in () could never be generated.
A neo-constructionist perspective on argument structure
The table below summarizes again the correspondences between the syntactic
position of arguments (DPs or roots) and their interpretation:
() Syntactic positions and semantic interpretation
... Against root ontologies I argue that roots must be treated on a par with
DP arguments. That means that roots, as DPs, receive a particular interpretation
depending on their position in the structure. For instance, a root like HAMMER may be
interpreted as an Effected Object (see ()), Terminal Ground (see ()) or Co-event
(see ()), depending on the configuration where it is merged:
() Sue hammered (the metal) for hours.
[VoiceP Sue [Voice’ Voice [vP [pP p the metal] [vP v HAMMER]]]]
() Sue hammered the metal in five minutes.
[VoiceP Sue [Voice’ Voice [vP [PathP the metal [Path’ Path [PlaceP the metal [Place’
Place HAMMER]]]]]]]
() Sue hammered the metal flat.
[VoiceP Sue [Voice’ Voice [vP [v v HAMMER] [PathP the metal [Path’ Path [PlaceP the
metal [Place’ Place FLAT]]]]]]]
In (), the root HAMMER is understood as an Effected Object, since it is the comple-
ment of v; as such, it describes the result of an activity and, accordingly, is compatible
with an atelic reading of the predicate. In (), the root is understood as a Terminal
Ground, since it is embedded in a PathP. Therefore, it depicts a final state (‘the state
of being hammered’), which, accordingly, habilitates a telic interpretation. Finally, in
() the root is interpreted as a Co-event by virtue of its being merged as an adjunct
to v: it specifies the way in which the event, here an externally originated change of
state, takes place.
The present framework
Assuming that roots are freely merged as arguments—again, with the proviso
that they are excluded from specifier position—root ontologies, that is, classifica-
tions of roots according to the possibilities they display of being inserted in the
structure as based on their semantic properties, turn out to be just a descriptive
artefact. Root ontologies are assumed in works such as Harley (), Levinson
(, ), and Rappaport Hovav and Levin (). Harley (), for instance,
proposes that instrument-naming verbs, such as hammer or rake, involve a root
that names an instrument (a hammer, a rake), and so the root is not merged in an
argumental position, for instance, in a position where it is interpreted as a final
state. Following Mateu and Acedo-Matellán (), and Acedo-Matellán and Mateu
(), I claim instead that, if HAMMER or RAKE name an instrument, that fact clearly
belongs to the encyclopaedic semantics and, hence, cannot determine where in the
structure the root is merged. Thus, for instance, Acedo-Matellán and Mateu ()
argue that instrument-naming roots can identify a resulting state in a change-of
state predicate. Verbs such as brush or rake can readily be used with depictive
secondary predication, which, according to Rapoport () and Mateu (), is
only compatible with change-of-state predicates, and not with activity predicates like
those headed by push:
() Acedo-Matellán and Mateu (: –)
a. Don’t brush the coat wet or you’ll ruin it.
b. He raked the field dry.
c. ??He pushed the mare pregnant.
In addition, roots like BREAK, which very intuitively name a result and not a manner
(Rappaport Hovav and Levin ; Rappaport Hovav and Levin ), can none-
theless be used as encoding manner co-events. For instance, in the following example
there is no entailment that the hammer head actually broke, so the verb break cannot
be said to encode a result state predicated of any overt participant in the event:
() McIntyre (: )
The hammer head broke off.
I assume that in the above example the root BREAK has been merged as an adjunct to v,
whereby it is interpreted as a co-event of the main eventuality of the hammer head
separating from the hammer (see section . for more examples with break).
Facts such as these suggest that the interpretation of a root as encoding Co-event,
(Terminal) Ground, or Effected Object depends on where it is merged in the
structure, rather than on a deterministic marking.18 See sections ... and ..
for more related discussion.
18
For further theoretical and empirical arguments against root ontologies and, in fact, against any
diacritic marking on roots, see Acquaviva (). See also Marantz (), who shows that the impression
A neo-constructionist perspective on argument structure
of the existence of different classes of roots boils down to (contextually determined) allosemy. Finally, see
Schäfer (: ) for the position that the suitability of a given root for the alternants of the Causative
Alternation depends on its conceptual properties, rather than on any grammatical marking.
The present framework
19
Path is of course not completely equivalent to Borer’s (b) AspQ: on the one hand, AspQP, though
entailing a measured change, does not entail the interpretation of a final location/state. On the other hand,
Borer contends that although in some languages the only way to license AspQ is by merging a DP
conveying a definite quantity as its specifier, in some other languages/constructions AspQ is argued to be
licensed independently, through particles, for example (see Borer b, chapters and ).
A neo-constructionist perspective on argument structure
Sue Voice’
Voice vP
v = put PathP
Place on
This dissociation motivates providing different structural positions for the Meas-
urer, the Figure, and the Ground, and positing movement to Spec-Path to explain
why a single DP can be simultaneously interpreted as Figure and Measurer or
as Ground and Measurer. In turn, observe that the projection of PathP and,
hence, the possibility that the predicate is telic depends, crucially, on the projec-
tion of a PlaceP. This state of affairs, together with the assumption that
telicity arises when a quantity DP from PlaceP is merged as Spec-Path, naturally
accounts for the fact that in intransitive telic predicates endowed with an argu-
mental PP (PlaceP) the subject is a Subject-of-quantity, pace Borer (b) (see
section ..).
Three possibilities arise as to the type of DP internally merged as Measurer and the
type of inner-aspectual interpretation yielded in conjunction with PathP: that the DP
is a quantity description (the books, some books, three books, etc.), a bare plural
(books), or a mass DP (paper):
The present framework
‘foal’ select the HAVE auxiliary in the perfect tense, like atelic activity verbs like
cantare, and unlike run-of-the-mill intransitive telic verbs like arrivare ‘arrive’,
which select BE. As was pointed out already at the end of section .., this kind of
example ((), repeated below) is problematic for a theory like Borer’s (b), in
which telicity is always grammatically represented:
() Italian; Mateu (a)
La giumenta {ha figliato/ *è figliata} in/??per due ore.
the mare(F) has foaled.M.SG is foaled.F.SG in/for two hours
‘The mare has foaled in two hours’
Telicity here depends solely on the fact that the Effected Object refers to a bounded
entity (figlio ‘son’). See Harley () for much relevant discussion.
Atelicity can be claimed to emerge from a greater variety of situations in com-
parison with telicity. First, Effected Object predicates license an atelic interpretation,
when they are roots, quantity DPs, bare plurals, or mass DPs:
() Atelicity with Effected Objects
a. Sue danced for an hour.
b. Sue did a dance for an hour.
c. Sue did dances for an hour.
d. Sue did work for an hour.
Predicates with a single p-projection, PlaceP, and, hence, a Central Ground, are
atelic, since they cannot present the location/state as final or resulting. This atelicity
obtains independently of the quantificational properties of the DP merged as
Spec-Place:
() Atelicity with Central Grounds
a. {These people/People} have been in Barcelona for a day.
b. Sue has loved {Jane/tomatoes} for years.
A predicate involving a PathP may be compatible with durative adverbials in three
circumstances. The first has already been illustrated in ()b and ()c: a non-
quantity Measurer (books, paper) yields an atelic interpretation in which the
transition encoded by PathP is entailed to have been partly carried out but, since
the quantity denoted by the DP is not definite, the transition corresponding to the
whole event cannot be calculated and, hence, the event—or, rather, the description
thereof—cannot be telic. On the other hand, predicates with a PathP structure
may license a durative adverbial by virtue of their embedding a PlaceP, which, as
has been pointed out, establishes a predicative relation. In particular, a PlaceP
embedded within a PathP may license an interpretation in which the resultant
location/state is measured by the for-temporal adverbial (Binnick , cited by
Dowty : ):
The present framework
... Locality domains for special meaning With Marantz (, ), I claim
that the special meaning ascribed to either word-sized units or bigger units must boil
down to contextually determined special meaning for roots, and that those special
meanings, like any non-compositional meaning, are listed in the Encyclopaedia.
Indeed, on the one hand, the Encyclopaedia cannot store chunks of structure,
since, from a strictly derivational point of view, structure cannot be stored (see
section .. for a critique of the l-/s-syntax difference in the same spirit); on the
other hand, structure cannot carry special meaning, since it depends uniquely on
A neo-constructionist perspective on argument structure
v PathP
Path PlaceP
calones Place’
Place ob
20
This is a claim made also for prefixes in the Slavic languages, particularly for so-called internal
prefixes, merged, by assumption, within the vP. See section .. for relevant examples and references.
The present framework
The semantic transparency of ()a is reflected in the analysis of (). The root OB
‘against’ is interpreted as a Terminal Ground, since it is embedded in a PlaceP in turn
embedded within a PathP structure: it depicts the final location of the Figure calones
‘soldiers’ servants’. I assume that the dative armatis ‘armed men’ is a possessive dative
understood as inalienably possessing the region identified by PlaceP—see section
.. for details on the so-called directional dative. The Figure calones raises to Spec-
Path, where it is interpreted as a Measurer for the event: the event is over when all the
calones ‘soldiers’ servants’ end up in front of the armed men (armatis). The predicate
is unaccusative, since Voice is not projected. To v is adjoined the root CURR, which
specifies the way in which the change of location takes place (running). On the other
hand, ()b is not less transparent than ()a, and it receives a similar analysis:
() Analysis of ()b
vP
v PathP
Path PlaceP
haec Place’
Place ob
I claim that the structural semantics of the verb in ()b is the very same as that in
()a. It could not be otherwise, since the meaning inherent to syntactic configur-
ations simply cannot be overridden. Both describe a telic change of state/location.
However, since the roots CURR ‘run’ and OB ‘against’ find themselves within the same
local domain for interpretation, they can trigger special meanings for each other. In
particular, the Encyclopaedic entries of both CURR and OB specify that a special
metaphorical meaning may be triggered in the presence of each other. Possibly
CURR ‘run’ is bleached out into conveying something like suddenness, while OB
‘against’ is reduced to a deictic marker. The Encyclopaedia need not specify the
extension of the domain within which that special meaning may be triggered: that is
provided by the syntax. Specifically, both roots are ‘visible’ to each other if and only if
they fall within the same Spell-Out domain.21
21
Interestingly, as pointed out by García Hernández (: –, : ), among others, the
conceptual meaning of verbal prefixes in Latin is different—more conservative—from that of their
corresponding prepositions. Thus, while the prefix de- conveys a downward orientation (García
Hernández , ) the preposition de can be paraphrased as ‘from’ or ‘away from’. This fact would
be easily accommodated in a theory in which the prefix and the verb—and not the verb and the
preposition—find themselves in an environment that is local enough for them to trigger a special meaning
in each other.
A neo-constructionist perspective on argument structure
. Summary
In this chapter I have made explicit my assumptions on the nature of the lexicon-
syntax interface. I have begun by introducing a fundamental distinction between
endo-skeletal and exo-skeletal theories. The former propose that the syntactic and
semantic properties of linguistic expressions are but a projection of lexical items,
while for the latter they emerge, largely, from the structure itself, lexical items being
reduced to conveyors of grammatically opaque, encyclopaedic content. After intro-
ducing the seminal work of Hale and Keyser I have revised three neo-constructionist
models: Mateu’s () theory of the relational syntax and semantics of argument
structure, Borer’s (b) syntactic theory of event structure, and the DM version of
the Minimalist Program for the architecture of grammar. I have then presented a
neo-constructionist model in which argument/event structure configurations are
created in the syntax through the application of free Merge. Structure is created on
the functional heads v, Place, and Path. Roots and DPs are merged in argumental
positions, a circumstance derived from an abandonment of the l-/s-syntax distinction
of the Halekeyserian model. Roots and DPs receive an argumental interpretation
according to the position that they occupy in the structure. Crucially, roots cannot
project structure, unlike some implementations of the DM model. As in any other
Minimalist account, the structures generated by the syntax are interpreted at the
interfaces. As far as semantic interpretation is concerned, I have emphasized the
distinction between structural semantics, emerging from the structure, and encyclo-
paedic semantics, encapsulated in the roots. I have also paid attention to the
aspectual interpretation of configurations, establishing that a Path projection is
responsible for a telic interpretation of the event if a quantity DP is merged at its
specifier.
3
tells us otherwise. For instance, as shown in (), the Latin conjunction -que ‘and’
encliticizes to the word on its left and triggers stress shift, revealing that the whole
string is behaving like a phonological word:
() Latin; Nespor and Vogel (: ), in Julien (: )
virum [ˈwi:ɾum] / virum=que [wi:ˈɾumkwe]
man.ACC.SG man.ACC.SG=and
Thus, virumque behaves prosodically in exactly the same way as any other word of
more than two syllables where the penultimate syllable is heavy. However, on no
sound syntactic account could -que and the host be analysed as one and the same
syntactic atom. Out of the domain of clitics, situations exist where arguably the same
components can be found within a phonological word or distributed in different
phonological words, depending on the context, as those italicized in the following
pairs of sentences:
() Marantz (), in Newell (: )
a. John cried.
b. Did John cry?
These are some of the very numerous cases of the indirect relation between
prosodic words and syntactic atoms. In this vein, I defend the view that phrases
interact syntactically and semantically with sub-word units, in consonance with
one of the main tenets of DM: Syntactic Hierarchical Structure All the Way
Down. In Harley and Noyer’s (: ) words, this tenet ‘entails that elements
within syntax and within morphology enter into the same types of constituent
structures (such as can be diagrammed through binary branching trees)’. As
was claimed in Chapter , DPs and roots may both occupy argumental posi-
tions in the structure. Observe the predicates in (). It is argued that they
correspond to the same configuration and, hence, yield the same structural
semantics:
v (= i/GO) PathP
Marcus Path’
Path = to PlaceP
Marcus Place’
Place ex/out
Specifically, the same predicative relation is claimed to hold between the unaccusative
subject Marcus and the locative pieces ex- and out. However, the morphophono-
logical packaging of the material is different in ()a and ()b: while the sequence
ex- ends up prefixed to the verb in Latin, its English counterpart out remains an
independent word in English. These facts support a view in which words are the
result of a variety of packaging mechanisms at PF operating on the representation
yielded by the syntax. Since the application of these mechanisms meets morpho-
phonological requirements of the nodes, cross-linguistic variation can be reduced to
how those nodes are phonologically specified.
() Latin
a. amo ‘I love’ / amavi ‘I have loved, I loved’
b. hortor ‘I order’ / hortatus sum ‘I (masc.) have ordered, I ordered’
Embick () further demonstrates that deponency is orthogonal to argument
structure and lexical semantics. Thus, for instance, hortor, in spite of its exclusively
passive morphology, appears in both transitive (see ()a) and passive sentences
(see ()b):
() Latin; Caes. Civ. , , and Varro in Prisc. GL. II, , (in Embick : )
a. Regemque hortatus est, ut [ . . . ] legatos.
king.ACC.SG=and order.PRF.SG.M that ambassador.ACC.PL
ad Achillam mitteret.
to Achilla.ACC send.IPFV.SBJV.SG
‘And he ordered the king to send ambassadors to Achilla.’
b. Ab amicis hortare-tur.
by friend.ABL.PL urge-IPFV.SBJV-PASS.SG
‘He was urged by friends.’
1
See Borer () for arguments against the indexical, non-phonological notation of roots. See De
Belder (); Acedo-Matellán and Real-Puigdollers (); and De Belder and van Craenenbroeck ()
for late-insertion theories of roots.
The syntax-morphology interface
Following Embick (: ff.), I assume that a cyclic domain headed by cyclic
node x is composed of x, the complement of x, and the set of all non-cyclic heads
(W, Z . . . ) between x and the immediately higher cyclic head y (I represent cyclic
domains within braces):
() y {W Z x COMPL}
According to the same author, Spell-Out proceeds in the following fashion: when
cyclic head y is merged, the lower cyclic domain headed by x is spelled out. For
instance, any DP (a cyclic domain) within the complement of v (a cyclic head) is
spelled out upon merger of v. This is the reason why DPs are phonologically
processed independently of the rest of the material in the vP.
. Operations at PF
.. Morphological Merger
Within DM a range of operations have been proposed to account for syntax-
morphology mismatches. Here I will concentrate on Morphological Merger
(Marantz , ). Marantz (: ) defines Morphological Merger as follows:
‘at any level of syntactic analysis (D-Structure, S-Structure, phonological structure), a
relation between X and Y may be replaced by (expressed by) the affixation of the
lexical head of X to the lexical head of Y’. This operation aimed at capturing cases
where morphemes appear dislocated from the position in which they are interpreted:
clitics, inflectional morphemes like causative affixes, or elements in words showing
so-called bracketing paradoxes (Pesetsky ). Embick and Noyer (, )
propose that Morphological Merger translates into somehow different ‘varieties’
depending on whether it applies before or after Vocabulary Insertion, that is, before
or after the insertion of the exponents of the linguistic expression. Assuming a Late
Linearization Hypothesis (Embick and Noyer : ), that is, assuming that the
set of terminal heads are provided with linear order and adjacency at Vocabulary
Insertion, it turns out that Morphological Merger is relativized with respect to
linearization itself: before Vocabulary Insertion it affects terminals that bear a purely
structural relation and after Vocabulary Insertion it affects terminals that bear a
linear and adjacency relation. In this work I will refer to the ‘variety’ of Morpho-
logical Merger that takes place before Vocabulary Insertion. An illustration thereof is
Embick and Noyer’s treatment of the presence of inflection (T) in the verb in English.
Independent evidence shows that the verb does not move to T in this language, in
overt syntax. To account for the surface adjacency of T and v, Embick and Noyer
(: ) propose that Morphological Merger lowers T to v, which is the head of
T’s sister, vP:
() Mary [TP t [vP loudly play-ed the trumpet.]]
Operations at PF
This example shows that Morphological Merger, when applying before Vocabulary
Insertion, is only sensitive to structural contiguity and not to linear adjacency:
although the vP adjunct loudly linearly intervenes between T and v, it does not
block Morphological Merger. However, when a different head appears between the
two relevant heads, an intervention effect does block Merger (Embick and Noyer
: , footnote ). Thus, if NegP projects between T and vP, T cannot move to
v, and do-insertion must take place in order to license the T node:
() *Mary [TP t [NegP not [vP play-ed the trumpet.]]]
() Mary [TP did [NegP ’nt [vP play the trumpet.]]]
Both in Marantz () and in Embick and Noyer (, ), Morphological
Merger is illustrated with cases in which, given a head X and its phrasal sister YP, X is
affixed onto Y, the head of YP, as illustrated above. This is called Lowering in Embick
and Noyer (, ):
() Lowering of X0 to Y0 (Embick and Noyer : )
[XP X0 . . . [YP . . . Y0 . . . ]] v [XP . . . [YP . . . [Y Y0þX0] . . . ]]
However, in Marantz’s original formulation of Merger there is no restriction as to the
‘upward’ or ‘downward’ sense in which Merger should operate. By virtue of this
formulation the relationship between XP and YP could very well be traded by the
raising of Y to X, that is, by ‘PF head movement’:
() PF Raising of Y0 to X0
[XP X0 . . . [YP . . . Y0 . . . ]] ! [XP [Y Y0þX0] . . . [YP . . . ]]
In this work I will adopt Raising as a variety of Morphological Merger before
Vocabulary Insertion.2 Raising can be illustrated with the derivation of a verb of
change of state like Catalan aparèixer ‘appear’. After Spell-Out, there is a successive
movement of the terminal nodes, beginning with the root, up to v (I am ignoring here
the functional heads above v):
2
Assuming Raising at PF evidently brings us to the still unsettled debate of whether there is also head
movement in the syntax proper. Chomsky () has argued that there is not, mostly on theoretical
grounds: head movement does not respect the Extension Condition (Chomsky ), that is, it does not
target the root of the derivation. Moreover, it does not seem to trigger semantic effects, which I find is the
most compelling empirical argument against the view that head movement takes place in the syntax. See
Matushansky (); Roberts (); and Bauke () for arguments in favour of head movement having
to take place in the syntax.
The syntax-morphology interface
v PathP
Path’
Path Place
Place
Place apareix
b. Successive Raising to v
v
Path v
Place Path
apareix Place
c. Linearization
APAREIX-Place-Path-v
d. Vocabulary Insertion
apareix-∅-∅-∅
See section .. for details on linearization and Vocabulary Insertion.
at hand. In later chapters I provide case studies of such failure in the abovementioned
languages. I will argue that the functional node Path in languages like Romance
languages, Latin, Slavic, and Ancient Greek is exponent-defective in this sense. For
instance, in Romance languages the Path node has a limited set of Vocabulary Items all
of which establish, as a condition for the insertion of the exponent, that Path be strictly
adjacent to the node v (e represents here different exponents):
() Path ⟷ e / _-v
In these languages Path does not have an elsewhere exponent, i.e., an exponent
insertable independently of any insertion frame, as it does in English (i.e., to). Hence,
if Path is not adjacent to the node v when Vocabulary Insertion takes place, it will not
receive any exponent, leading to a failure in the interpretation of the structure. The
theory developed here is, therefore, a crashing theory. Importantly, contemplating
crashes at PF, that is, the possibility that some structures built by the syntax cannot be
lexicalized, is not a hallmark of DM, in which, as pointed out above, for a given node
F there is always a default Vocabulary Item that can insert an exponent (frequently a
null one) in F where no other can. However, something like a PF crash is assumed in
theories like Borer’s (a, b, ), in which it is argued that cross-linguistic
variation is also reduced to the (un)availability of particular phonological represen-
tations of functional nodes or combinations thereof (see, for instance, Borer b:
). Similarly, in Nanosyntax, it is assumed that languages only differ in what lexical
representations they have available for a particular freely built syntactic representa-
tion (Fábregas ; Ramchand and Svenonius ). The notion of PF crash based
on an uninterpretable exponent-defective node can be modelled through Fábregas’s
(: ) Exhaustive Lexicalisation Principle, which states that ‘Every syntactic
feature must be lexicalised.’
According to the insertion frame of this Vocabulary Item, at the time of Vocabulary
Insertion, Path must be strictly left-adjacent to v, that is, a prefix. This can be brought
about through an instance of Raising bringing Path onto v before Vocabulary
Insertion:
v PathP
Path v PlaceP
Linearization takes place after Raising, effecting the sequence Path-v. Importantly, if
Raising does not take place, or if some overt node linearly intervenes between Path
and v, Vocabulary Insertion into Path cannot be effected, since the contextual
requirement is not met and Path is exponent-defective, that is, it does not have an
elsewhere Vocabulary Item to rescue its phonological interpretation. The realization
of Path in these languages (and also in Basque, Japanese, Hebrew, etc.) seems to be
intimately dependent on that of v. I argue that this is the reason behind the well-
known awkwardness of predicates like the following one in Romance (Talmy ,
; Mateu and Rigau , among many others):
() Catalan
*Ella ballà a l’habitació. (In the directional sense.)
she danced at the=room
The syntax-morphology interface
In the above predicate, the root BALL ‘dance’ is sister to v in a complex v head,
providing the specification of a co-event (in this case, a dancing event accom-
panies the main event of getting into the room; see section ...). I assume, as
does Embick (), that roots raise to the immediately upper functional category
(see section . for more extended discussion). This implies that after Path has
raised to v and the whole structure has been linearized, the root linearly inter-
venes between Path and v. Importantly, as argued by Embick () and Marantz
(), an element does not linearly intervene when phonologically null. It is not
the case here, however, since the root BALL is phonologically overt. As a result,
the contextual specification of Path is not met, and PF fails to interpret the
structure.
A similar, though not identical effect can be found in languages like Latin, Slavic,
or Ancient Greek. In these languages, too, the Path head is exponent-defective,
although in a different way. I assume that one of the Vocabulary Items for Path in
these languages reads as follows (see section . for a revision of this Vocabulary
Item):
v PathP
v equit Path’
Path PlaceP
Place’
Place ad
b. Raising
v
Path v
ad Place
c. Linearization
AD-Place-Path-EQUIT-v
d. Vocabulary Insertion3
ad-∅-∅-equit-∅
The raising of the terminal nodes onto v, to form a complex head, provokes the
prefixation effect typical of languages like Latin or Slavic, and this will be amply
discussed in later chapters. Importantly, as will be shown in Chapter , what is not
found in these languages is equivalent constructions in which the final location is not
prefixed onto the verb:
() Latin; made-up example
#Caesar ad portas equitavit.
Caesar.NOM at gates.ACC ride.PRF.SG
‘Caesar rides towards the gates.’ / *‘Caesar rides up to the gates.’
The above example is not ungrammatical in Latin, but it does not convey the resultative
reading in which Caesar actually arrives at the gates (on horseback)—see Van der Heyde
() for an early observation along these lines. At most, the PP ad portas ‘towards the
3
Here and elsewhere I ignore thematic vowels (here a) in Latin and Romance. Following Oltra-Massuet
() and Oltra-Massuet and Arregi (), I assume that thematic vowels are introduced postsyntacti-
cally (see Embick for the notion of dissociated morpheme). The issue is tangential to the current
discussion.
The syntax-morphology interface
gates’ is interpreted as a directional adjunct, not as entailing the final location. Thus, this
example cannot involve the structure in ()a. The reason for the failure of the
resultative interpretation, I argue, is that the prefixation requirement on Path, described
above, cannot be met. The only derivation compatible with the sentence is one in which
ad portas ‘towards the gates’ is an adjunct to an activity vP involving the root EQUIT ‘ride’:
() Ad portas equitare ‘ride towards the doors’.
vP
pP vP
p portas v equit
p ad
Since Raising is defined in structural terms, and applying from one head to the
immediately higher c-commanding head, it cannot target anything within the
adjunct PP. This in turn explains why in motion constructions involving a prefixed
verb, such as that in (), the adjunct reading of the prefix (in this case a ‘towards’
reading) is out, and the only interpretation possible is that of a change of location.
In Chapters , , and we will have more opportunities to see how the morpho-
logical interpretation of syntactic structures gives rise to observable systematic cross-
linguistic differences within the realm of argument structure.
. Summary
In this chapter I have presented a (partial) theory of the PF-interpretation of linguistic
structure. It is this interpretation, I argue, that cross-linguistic differences are restricted
to. I have assumed that, unlike the syntax-semantics interface, the syntax-morphology
interface can be non-isomorphic. PF computation operates in cycles defined by the
categorizers like v and a. Importantly for the main discussion in Chapter , the material
computed in a given cycle is not available for another cycle. I have discussed a series of
PF operations. A Raising operation, taking place before Vocabulary Insertion, brings
the nodes together, yielding complex words. This operation applies freely, its results
being filtered by the exponents available at Vocabulary Insertion, which takes place
after linearization. Crucially, I have argued that some functional heads in certain
languages are exponent-defective, that is, they do not count with an elsewhere
Vocabulary Item when the rest of Vocabulary Items do not meet the contextual
conditions of insertion. In these cases, if Vocabulary Insertion cannot take place, the
derivation crashes at PF. Differences in the specification of Vocabulary Items for
functional heads are then at the basis of the explanation of patterns of cross-linguistic
variation in argument structure, as will be shown in subsequent chapters.
4
In this chapter I use the theoretical tools introduced in Chapters and to analyse a
wide range of argument structure phenomena in Latin. A quick glance at the
Dictionnaire Latin-Français by Gaffiot () shows that many composite verbal
lexical entries in Latin receive a periphrastic definition in French. Importantly, the
correspondence between the morphological components of the Latin verb and the
syntactic components in the Romance periphrasis appears to be systematic. The
following entries involving the prefix ex- ‘out’ illustrate the fact:
() Latin; Gaffiot ()
a. ex-cutio
out-shake.SG
‘Faire sortir ou tomber en secouant’ (‘make go out or fall shaking’)
b. ex-cudo
out-beat.SG
‘faire sortir en frappant’ (‘make go out beating’)
c. e-repo
out-crawl.SG
‘sortir en rampant, en se traînant’ (‘go out crawling’)
In the above examples, the prefix ex- ‘out’ (with the form e- in erepo), seems to
correspond, in the French translation, to a whole verb, namely (faire) sortir ‘(make)
go out’, while the semantic content of the simple verb in each case is translated as a
manner adverbial (en secouant ‘shaking’; en frappant ‘beating’; en rampant, en se
traînant ‘crawling’). For motion events in general, while Latin expresses the trajec-
tory and final location within one morpheme and the ‘kind’ of motion—shaking,
beating, and crawling, respectively, in ()a to ()c—within a different morpheme
(namely, the verb: quatio ‘shake’, cudo ‘beat’, repo ‘crawl’), French lexicalizes the
trajectory and final location in the form of an independent and monomorphemic
verb—as sortir ‘go out’, entrer ‘go in’, etc.—and the kind of motion is conveyed by an
optional adjunct. This difference in the expression of the components of a (motion)
event shown by Latin and French actually corresponds to a typological difference
claimed by Talmy () to divide many of the world’s languages into two blocks:
satellite-framed languages (Latin-like languages) and verb-framed languages (French-
like languages). After introducing Talmy’s insightful observations on the cross-
linguistic expression of events of change, I model his theory in terms of the one
put forward in Chapter . It will be argued that cross-linguistic differences are purely
morphological and, as such, derive from the language-specific morphological speci-
fication in the Vocabulary Items of functional nodes. In section . I describe the
possible morphosyntactic manifestations of PathP in Latin. The bulk of the chapter is
devoted to showing the validity of Talmy’s (: ) observation that Latin is a
satellite-framed language. I explore and analyse, to that end, a set of constructions
involving change or transition (in my terms, a PathP).
specifies that the Ground is the starting point of movement, VIA, which signifies that
the Ground is something located in the Path, but which is neither the starting point
nor the end point, etc. In () the Vector is TO, and is codified in the -to morph of
into, while in () the Vector is AT, and lies in the preposition in.
The Conformation creates a geometrical shaping of the Ground, which comes then
to be conceptualized as a volume, an enclosure, a plane, etc. The conformation in
both () and () is the one corresponding to an enclosure, and could be paraphrased
as INSIDE. Note that, in both cases, it is expressed by the preposition in, which in ()
encodes, in addition, the Vector AT, and in () is morphologically attached to the TO
Vector encoder -to. A volume conformation, which we could dub SURFACE, applied
to the same motion event, could yield The cat walks onto the hat and There stood a cat
on the hat, respectively. I will argue that the different interpretation of the Conform-
ation is not grammatically represented, and that it is instead encoded as a choice of
different roots.
The Deictic component conveys whether the sense of the Path is towards the
speaker or away from the speaker. The verbs to come and to go exemplify, respect-
ively, a þSPEAKER (towards the speaker) and a -SPEAKER Deictic.1
A last important element must be mentioned, which, although not itself a com-
ponent of the motion event, is very often associated with it. It is what Talmy calls the
Co-event, that is, an event that is related in some way to the Motion event, which is
considered, in turn, the Framing event. That relation can be of different types:
causation, manner, etc. In the case of () and (), the Co-event expresses manner,
more specifically, the way in which the movement or the stationariness takes place,
a walking event in () and a standing event in (). Note that in both sentences this
Co-event is expressed via the verb (the root of the verb), together with the Motion
component, MOVE and BEAT, respectively.
Having put forward the main elements involved in the expression of motion,
I now want to introduce the major cross-linguistic difference referred to in the
introduction to this chapter. Talmy (, : –) proposes that languages can
be ascribed to groups in which there is a systematic encodement, in a single
morpheme, of the same components of a motion event. Specifically, he focuses on
the Core Schema, and describes two possibilities as to its surface (syntactic) expres-
sion: the Core Schema can be expressed within the verb, conflated—that is, fused
into the same piece—with the motion component, or it can be expressed through
an independent element of the predicate that he calls satellite, ‘the grammatical
category of any constituent other than a noun phrase or prepositional-phrase
complement that is in a sister relation to the verb root’ (Talmy : –).
Languages that primarily opt for the first way of encoding the Core Schema are
1
The technical names INSIDE, SURFACE, þSPEAKER, and -SPEAKER are creations of my own
(Talmy : refers to þSPEAKER as hither and to -SPEAKER as hence).
Latin as a satellite-framed language
called verb-framed languages, while languages that choose the second way are called
satellite-framed languages.2 What is of relevance to the present discussion is that
there is a kind of complementary distribution between the expression of the Core
Schema and the expression of the Co-event, such that in v-framed languages the Co-
event is not conflated in the verb, and usually appears in an adjunct phrase, while in
s-framed languages the Co-event can be readily expressed within the verb, as is the
case with the manner Co-event in () and () above. Although we have already seen
how an s-framed language distributes the Core Schema and Co-event components,
in () and (), let us now introduce a minimal pair involving Catalan (a v-framed
language) and English (as pointed out already, an s-framed language) expressing a
motion event with a manner (of motion) Co-event:
() Catalan and English
a. La pilota va [entrar]verb: Motion+Core Schema
the ball PST.SG go_in.INF
[rodolant.]adjunct: Co-event (manner)
rolling
b. The ball [rolled]verb: MotionþCo-event (manner) [in.]satellite: Core Schema
As glossed in the examples, the Catalan sentence expresses the trajectory of the ball
(the Core Schema, here equivalent to a trajectory ending up in some enclosure)
within the verb, while the manner in which it moves along that trajectory is encoded
in an independent and optional gerund phrase. In English, those same components
of the motion event are expressed in a different way: the Core Schema is separated
from the verb and is expressed as a satellite, while the manner Co-event is fused
together with the Motion within the verb. This different morphosyntactic structuring
of the motion event is correlated, as Talmy (: ff.) observes, with certain facts
about the lexicon of each type of language. For instance, Catalan (and, in general,
v-framed languages) has a great variety of roots expressing directed motion at its
disposal, each corresponding to a particular Core Schema component, while English
lacks many of those specialized verbs: Cat. entrar, ‘go in’; sortir, ‘go out’; treure, ‘take
out, off ’; ficar, ‘put in’; etc.
2
See Talmy (: –, –) for another major typological group of languages, namely languages
in which it is the Figure component that gets lexicalized into the verb.
Talmy’s (2000) theory of change events
the final, resultant state, the Motion component is to be identified with the event
itself, and the Co-event is the way in which the change of state takes place. The
examples from German and Spanish following illustrate, respectively, how s-framed
and v-framed languages express events of change of state:
() German and Spanish; Talmy (: )
a. Der Hund hat [den Schuh]Figure [kaputt]Core schema -[gebissen.]Event+Co-event
the dog has the shoe in_pieces bite.PTCP.PST
‘The dog bit the shoe to pieces.’
b. El perro [destrozó]Event+Core schema [el zapato]Figure [a mordiscos.]Co-event
the dog destroy.PRF.SG the shoe to bites
.. An asymmetric difference
As can be shown through a comparison of s-framed English and v-framed Catalan,
the s-/v-framed distinction happens not to be symmetric, that is, it does not yield two
groups of opposing languages. The asymmetry appears to consist in a wider avail-
ability of the v-framed strategy, which is allowed in typically s-framed languages
like English (Mateu ). The s-framed pattern, on the other hand, is precluded in
v-framed languages like Romance. Thus, English does have directional verbs, which,
not surprisingly, are mostly Latinate: to enter, to exit, to remove, etc. It can also
express events of change of state within a verb, as in The wind cleared the sky, The sun
melted the snow, etc. The opposite, however, is not found in v-framed languages:
while they can make use of the v-framed strategy, they cannot make use of the
s-framed strategy. Hence, typically s-framed constructions involving the expression
of a Co-event within the verb are ungrammatical in these languages:
() Catalan
*El vent bufà el cel clar.
the wind blew the sky clear
‘The wind blew the sky clear.’
() Catalan
*En Joan martellejà el metall pla.
the Joan hammered the metal flat
‘Joan hammered the metal flat.’
In section ... I provide a morphological analysis of this asymmetry.
like the following (see Hoekstra and Mulder for French and Torrego for
Spanish):
() Catalan; Mateu (: )
En aquesta coral n’hi canten molts, de nens.
in this choir PARTVE=LOC sing..PL many.PL of child.PL
‘There are many children who sing in this choir.’
According to Rigau (), in predicates such as (), the verb canten ‘they sing’ bears
an existential stative meaning close to that found in there-existential sentences.
Hence, a good paraphrase for () is the English translation provided underneath.
On the other hand, and according to Mateu (: ff.), there is evidence that
the construction is of unaccusative nature, as hallmarked by the possibility of
en-extraction (see () itself), and the licensing of postverbal bare plural subjects, as
shown below:
() Catalan
En aquesta coral hi canten nens.
in this choir LOC sing.PL children
‘Children sing in this choir.’
It is also telling, in this respect, that in Italian these constructions resist HAVE-
selection when put in the perfect:
() Italian; Centineo (: –), in Mateu (: )
??Ce ne ha nuotato molta, di gente, in quella piscina.
LOC PARTVE has swum many of people in that swimming pool
Importantly, Mateu (: ) highlights Centineo’s (: , footnote ) obser-
vation that, upon elicitation, some native informants attempted to use essere, the BE
auxiliary, in examples like (). I will assume with Mateu () that this type of
construction is unaccusative. I will analyse them as such, and I will explain why they
are fine in v-framed languages in section ....
3
See Bouchard () for a similar analysis of French movement verbs such as venir ‘come’ or aller ‘go’.
Latin as a satellite-framed language
v PathP
vMotion danceCo-event Path = to PlaceP
SueFigure Place’
() Catalan
En Joan eixí.
the Joan went_out
[vP [v vMotion ] [PathP Path [PlaceP En JoanFigure [Place’ Place EIXGround ]]]]
As for the Vector component, I shall assume that, at least when PathP appears
embedded within vP, it is always of value TO. In that sense the head Path is
significantly different from Talmy’s Path: it instantiates a transition into a final
location or state. In other words, Core Schemas are always goals, and not sources.
In a predicate such as She danced out of the room, hence, out of the room
corresponds to a goal of motion, describing where the dancing event will end
up. There is evidence for this position. For instance, change-of-state predicates
always describe a final, resultant state, and not an initial or medial state. In the
same way, there is no verb in any language that I know of lexicalizing the meaning
‘stop being’, which would correspond exactly to a predicate of change of state
expressing the source state. This is partly illustrated by the following paradigm
from Gehrke (), where turn must appear with a goal PP and cannot appear
with a source PP alone:
... A morphological account of the s-/v-framed difference The main aim of this
book is to explain cross-linguistic variation as depending on idiosyncratic morpho-
logical properties of functional heads. The s-/v-framed distinction will be dealt with
from this post-syntactic perspective. This means that the syntactic construction
of events of change, which are the locus of the distinction, and their interpretation
at LF are common to all languages, and that it is how those structures are
interpreted morphologically, at PF, which can vary from language to language.
I introduce the discussion in this chapter, although it will be of great importance in
Chapters and .
Observationally, the s-/v-framed distinction has to do with how morphs, in the
structuralist sense of the term, relate to morphemes, as Talmy’s definition of confla-
tion suggests: in s-framed constructions the same morph corresponds to (or con-
flates) the Motion and the Co-event components (here, v and a root adjoined to it,
respectively). In v-framed constructions the same morph corresponds to the Motion
and the Core Schema (here v and PathP). Since we know that s-framed languages
admit the v-framed strategy, but v-framed languages do not admit the s-framed
strategy (see section ..), there has to be a more restrictive mechanism in v-framed
languages than in s-framed ones, accounting for this asymmetry. Using the theoret-
ical tools introduced in Chapter , I propose that in v-framed languages like
Romance, the Path and v must be strictly adjacent to each other, which is accom-
plished through a Vocabulary Item like the following (e stands for exponent):
() Path ⟷ e / _-v
Assuming that Path is exponent-defective in these languages, that is, that it does not
have an elsewhere Vocabulary Item (a Vocabulary Item with no insertion frame), this
head can only receive an exponent if the contextual condition expressed in the above
Vocabulary Item is satisfied. In order to meet the insertion condition, Path must raise
to v in v-framed languages:
() Path-to-v Raising in v-framed languages
vP
v PathP
Path v PlaceP
In the case of predicates of change of state and change of location like Catalan eixir
‘go out’, in which all the nodes of the v-cycle raise to v, the Vocabulary Item for
Path establishes that it is realized with a null exponent when sandwiched between
Place and v:
() Path ⟷ ∅ / Place-_-v
Latin as a satellite-framed language
I illustrate with the derivation of En Joan eixí ‘Joan went out’. I ignore (here and
elsewhere) the terminal nodes higher than v and Voice that also belong in the vP
cycle (T, etc.):
() Catalan; PF-derivation of En Joan eixí
a. Structure delivered by syntax
[vP v [PathP [Path’ Path [PlaceP [Place’ Place EIX]]]]]
b. Raising
[v [Path [Place EIX Place] Path] v]
c. Linearization
EIX-Place-Path-v
d. Vocabulary Insertion
eix-∅-∅-∅
()a is the structure that arrives at PF. Raising forms a complex head out of the root,
Place, Path, and v. After linearization the nodes receive an exponent according to
their Vocabulary Items. I assume that Place and v also receive a null exponent in this
context. In particular, I assume that the following is one of the Vocabulary Items for
Place in v-framed languages:
() Place ⟷ ∅ / _-Path
As for v, its exponence may not be null, as pointed out by Oltra-Massuet ().
Thus, there are cases of predicates of change of state, involving the same structure as
in the above example, in which v is overtly realized:
() Catalan; non-null exponent for v
agud-itz-a-r
acute-ize-TH-INF
‘make more acute’
In these cases, the Vocabulary Item for v specifies an overt exponent in the context of
a particular root:
() v ⟷ itz / AGUD-v
Allomorphy of v can be triggered across Place and Path since these heads, although
linearly intervening, are null (see Embick ; Marantz ).
A slightly different derivation is involved in simple go-predicates involving a PP
like En Joan anà a la botiga ‘Joan went to the shop’:
() Catalan
En Joan anà a la botiga.
the Joan went at the shop
‘Joan went to the shop.’
Talmy’s (2000) theory of change events
the present account (as well as in Embick ; McIntyre ; Zubizarreta and Oh
; or Mateu b), typical s-framed constructions are analysed as involving the
adjunction of a root to v, being interpreted as a Manner Co-event. In v-framed
languages, this adjunction structure is not compatible with the Vocabulary Item
proposed above for Path. Assuming, as in Embick (), that roots must raise to the
upper functional head, the raising of the Co-event root to v, followed by linearization,
yields a sequence in which the root linearly intervenes between Path and v. Since the
root is overt, it blocks insertion of any exponent in Path. I illustrate below with the
derivation of ungrammatical *Ella ballà a l’habitació ‘She danced into the room’:
() Catalan
*Ella ballà a l’habitació. (In the directional sense.)
she danced at the=room
a. Structure delivered by syntax
[vP [v v BALL] [PathP [Path’ Path [PlaceP [Place’ Place {l’habitació}]]]]]
b. Raising
[vP [v Path [v BALL v]] [PathP [Path’ [PlaceP [Place’ Place {}]]]]]
c. Linearization
Path-BALL-v > Place
d. Vocabulary Insertion
?-ball-∅ > a
At Vocabulary Insertion, the strict adjacency condition imposed on Path is not met.
Since Path is exponent-defective, there is no other Vocabulary Item that can interpret
the Path node. This explanation can be extended to cover the disallowance of
complex AP resultative constructions in v-framed languages:
() Catalan; Mateu (: )
*El gos va bordar els pollastres desperts
the dog PST.SG bark.INF the chickens awake.M.PL
In this case the root BORD ‘bark’ provides the Co-event component by virtue of
its being associated with v. At Vocabulary Insertion it too intervenes between
Path and v, blocking the insertion of the exponent for Path. However, in the case
of AP resultative constructions with no independent co-event component (i.e. simple
AP resultative constructions; see section ...), the derivation is predicted to be
possible. Thus, in Romance languages we find AP resultative constructions with
causative verbs with no entailment of a co-event component:
() Catalan
El gos va deixar els pollastres marejats a lladrucs.
the dog PST.SG leave.INF the chickens dizzy.M.PL at barks
‘The dog got the chickens dizzy with its barking.’
Talmy’s (2000) theory of change events
In this case, I assume that the verb deixar ‘leave’ corresponds to a realization of Path
in the context of (active) Voice. Both v and Voice are realized with a null exponent:
() Path ⟷ deix / _-v-Voice
I also assume that the adjective is a PlaceP endowed with phi-features, and corres-
ponds to a PF cycle of its own (see section ..):
() Analysis of ()
a. Structure delivered by syntax
VoiceP
El gos Voice’
Voice vP
v PathP
Place marejat
b. Raising
[Voice [v Path v] Voice]
c. Linearization
Path-v-Voice
d. Vocabulary Insertion
deix-∅-∅
In s-framed languages, by contrast, the realization of Path is not necessarily
dependent on that of v. In English, for instance, there is a default Vocabulary Item
for Path, with an overt exponent:
() Path ⟷ to
Thus, a simple motion sentence like John went to the room is derived as follows (I am
ignoring, as usual, the presence of T, which of course triggers the allomorph wen):
() Derivation of John went to the room
a. Structure delivered by syntax
[vP v [PathP [Path’ Path [PlaceP [Place’ Place {}]]]]]
Latin as a satellite-framed language
b. Linearization
v > Path > Place
c. Vocabulary Insertion
wen(-t) > to > ∅
In English there is no strict adjacency condition on Path and v. Path may be realized
independently of v (above, in the absence of any Co-event root, realized as go/wen—
see Zubizarreta and Oh : ff. for discussion). The independence of Path and v
at Vocabulary Insertion makes it possible for a root, interpreted as Co-event, to
appear adjoined to v:
() PF-derivation of John tiptoed to the room
a. Structure delivered by syntax
[vP [v v TIPTOE] [PathP [Path’ Path [PlaceP [Place’ Place {}]]]]]
b. Raising of the root to v
[vP [v TIPTOE v] [PathP [Path’ Path [PlaceP [Place’ Place {}]]]]]
c. Linearization
TIPTOE-v > Path > Place
d. Vocabulary Insertion
tiptoe-∅ > to > ∅
On the other hand, nothing precludes, in a language like English, the generation of
predicates of change of state in which all heads of the vP cycle end up forming a
complex head:
() PF-derivation of The sky cleared
a. Structure delivered by syntax
[vP v [PathP [Path’ Path [PlaceP [Place’ Place CLEAR]]]]]
b. Raising
[v [Path [Place CLEAR Place] Path] v]
c. Linearization
CLEAR-Place-Path-v
d. Vocabulary Insertion
clear-∅-∅-∅
() presents the derivation of a v-framed construction in an s-framed language. It is
a v-framed construction, in Talmy’s terms, since the Core Schema is expressed within
the verb, and not independently of it. There is nothing in the morphophonological
specification of v or Path in English impeding the derivation of these cases. See
section . for more details.
The surface shape of PathP in Latin
Finally, recall from section .. that the requirement of a common phonological
realization of v and the Core Schema in v-framed languages is not effective when the
construction is stative, non-dynamic. In the present terms, this follows automatically
from the fact that the constructions in question do not feature a Path head. I illustrate
below with the analysis of (), repeated here as (), which follows the spirit of that
proposed by Hoekstra and Mulder () and Mateu (: ):
() Catalan
En aquesta coral hi canten nens.
in this choir LOC sing.PL children
‘Children sing in this choir.’
() PF-derivation of ()
a. Structure delivered by syntax
vP
v PlaceP
Place DEICTIC
b. Raising
[vP [v CANT V] [PlaceP [Place’ [Place DEICTIC Place]]]]
c. Linearization
CANT-v > DEICTIC-Place
d. Vocabulary Insertion
cant-∅ > hi-∅
In () the root is adjoined to v and is interpreted, consequently, as a Co-event. v is,
in this case, interpreted as a stative non-externally originated event, since neither
Voice nor Path is projected. The DP nens ‘children’ is a Figure and enters into a
predicative relation with an abstract deictic element merged as Compl-Place. Since
there is no Path head, there is no requirement for v to be linearly adjacent to it, and
the adjunction structure [v v CANT] may morphologically survive.4
4
In the analysis I am ignoring cliticization of hi onto T.
Latin as a satellite-framed language
or not. To sum up, directionality can be expressed through (i) a verbal prefix, (ii) a
PP, (iii) a combination of both prefix and PP, (iv) a combination of a prefix and a DP,
and (v) finally, and marginally, a (case-marked) DP. I will show that APs are not
possible encoders of the PathP, which is well attested in other s-framed languages like
Germanic. Finally, I will analyse how case is assigned in DPs and PPs in predicates of
directed motion.
VoicePass vP
v Path
Path PlaceP
cruor(e) Place’
Place abs
The surface shape of PathP in Latin
Flatus Voice’
Voice vP
v PathP
Place ex
The prefix originates as a root at Compl-Place. It is through Raising, at PF, that it ends
up pronounced in the same word as the verb (see section ..). In its original position
it is interpreted as a Terminal Ground, since PlaceP is embedded within a PathP. In the
case of (), for instance, the blood, cruore, ends up being off (the wound). On the
other hand, the root TERG, adjoined to v, is interpreted as Co-event: it is through wiping
that the blood ends up off the wound. Observe that in both cases I posit movement
from Spec-Place to Spec-Path, where the internal argument is interpreted as a Meas-
urer: the wiping and rolling events are over when the blood and the shrubs are off and
out, respectively. Recall from section ..., that Compl-Place is to be read necessarily
as a final location/state when PlaceP is embedded under a PathP projection. This might
seem counter-intuitive in the case of preverbs like ab(s)- ‘away’ or ex- ‘out’, which are
traditionally classified as ablative, that is, indicating a departing point (García
Hernández ). However, things become clearer when the distinction between
conceptual and structural meaning is seriously taken into account: EX, as a root,
expresses ‘outness’, a notion involving a reference point which, by inference, is taken
as the departure point—for instance, in () this departure point is the ground whence
the shrubs are uprooted; on the other hand, when embedded in a PlaceP, it comes to be
identified as a final location. So the final location is ‘outside the reference point’.
Analyses such as Molinari’s (: ), which do not acknowledge this difference
(or a syntactic theory of prefixation, for that matter), must resort to a double definition
for ‘ablative’ verbal prefixes such as ex-, namely, that they identify a departing point
and an arrival point that is outside the departing point.
In line with claims by Gehrke () for ambiguous prepositions in Germanic and
Real Puigdollers (, ) for ambiguous prepositions in Germanic and Romance,
I suggest that Latin verbal prefixes are never directional per se: the directionality is the
effect of their being merged as Compl-Place within a PathP. Evidence that this is the
right analysis is the fact that prefixes that may head directional, change predicates can
Latin as a satellite-framed language
also appear in stative predicates, combined with sum ‘be’. This is shown below, where
prefixes de- ‘away; down’ and ab(s)- ‘away’ are found in a transition predicate in the a
examples and in a stative, Pathless predicate in the b examples:
() Latin; Caes. Civ. , , and Ter. Phorm.
a. Ad naves de-currunt.
at ship.ACC.PL down-run.PL
‘They run down towards the ships.’
b. Argentum de-erat.
silver.NOM away-was.IPFV
‘Money was lacking.’
() Latin; Liv. , , and Plaut. Cas.
a. Inspectum vulnus
examine.PTCP.PFV.NOM.N.SG wound(N)NOM.SG
abs-terso cruore.
away-wipe.PTCP.PFV.ABL.M.SG blood(M)ABL.SG
‘That the wound had been examined after wiping the blood off ’.
b. Senex ab-est.
old_man.NOM away-is
‘The old man is missing.’
v PathP
Path Place
omnes Place
Place palatium
Place in
The surface shape of PathP in Latin
As shown in section .., there is evidence that a different analysis is preferable for
this kind of predicate. See also section .. for more considerations on the syntax of
directional PPs.
v PathP
v ced (tu) Path’
Path PlaceP
Path ad (tu) Place’
Place aedis
Place in
The root AD, by being merged as an adjunct to Path, is merely interpreted as directional.
The final location is specified by PlaceP in aedis ‘into the house’. As for cases in which the
prefix coincides with the preposition, Acedo-Matellán (, b) and Oniga (:
ff.) propose that the prefix and the preposition correspond to two pronounced copies
of a movement chain originating in the site of the preposition. This analysis cannot
account for the derivation of cases in which the prefix and the preposition do not
Latin as a satellite-framed language
coincide.5 If the preposition and the prefix are taken to be copies of the same object, it is
not clear why they should possess different phonological and semantic properties. On the
other hand, both this analysis and that put forward in () in section .. leave
unexplained the fact that PPs specifying final location in prefixed predicates are omissible
without the fundamental transition interpretation of the predicate being altered:
() Latin; Cic. Verr. , , ,
Subito ipse ac-currit.
suddenly self.NOM at-run.SG
‘Suddenly he himself arrives in haste.’
() Latin; Ov. Met. ,
Tergo velamina lapsa re-liquit.
back.ABL.SG veil.ACC.PL slip.PTCP.PFV.ACC.N.PL back-leave.SG
‘She left behind the veil which had slipped off her back.’
() Latin; Cato, Agr. ,
Tenuissimas radices ex-arabit.
slender.SUPERL.ACC.F.PL root.ACC.PL out-plough.FUT.SG
‘He will plough out the most slender roots.’
For these reasons I propose that the prefix originates at Compl-Place, and is therefore
interpreted as a Terminal Ground, and that the PP is in fact an adjunct to PathP or to
PlaceP, that is, a modifier of the direction component of the transition predicate or of
the final location. I exemplify with the analysis of ():
() Analysis of ()
vP
v PathP
v ced pP PathP
p aedis (tu) Path’
p in Path PlaceP
(tu) Place’
Place ad
5
See Lehmann () and López Moreda (). Interestingly, Biskup and Putnam () propose that
the German preposition/particle aus ‘out’ and the prefix ent- correspond to the same morpheme, spelled
out with different exponents according to its position:
(i) German; Biskup and Putnam (: )
Sie ent-steigt dem Auto. /Sie steigt aus dem Auto.
she ent-climbs the.DAT car she climbs out the.DAT car
‘She gets out of the car.’
This analysis does not seem adequate, however, for cases such as that in (), in which the prefix and the
preposition are not semantically equivalent.
The surface shape of PathP in Latin
6
According to Spencer and Zaretskaya (: ), in the Slavic languages, which, as I show in
Chapter , also make extensive use of verbal prefixes, PPs are omissible in predicates headed by a prefixed
verb—see, for Czech, Filip (: ), and for Russian, Rojina (: ). These authors too arrive at the
conclusion that in this kind of predicate the PP is an adjunct, the directional and resultative interpretation
stemming from the prefix itself.
Latin as a satellite-framed language
... Although these authors point these problems out, they do not provide a syntactic
analysis of this kind of predicate, and simply state that they correspond to lexicalizations,
opaque to the syntax. As we will see in sections to come, this cannot be the case, since
prefixed verbs, although they do not involve stranding, show systematic syntactic and
semantic properties which set them apart from their non-prefixed counterparts.
Taking into account the remarks above and the discussion in section .., which
made it clear that the transitional interpretation of the event depends exclusively on
the prefix, I propose that, at least for cases like (), the prefix originates at Compl-
Place, and that the ablative castris ‘out of the camp’ is an adjunct to PlaceP:
() Analysis of ()
Voice
pro Voice’
Voice vP
v PathP
pP PlaceP
p castris
omnes copias Place’
Place ex
7
With prefixed verbs the frequency of directional DPs grows considerably, as Hofmann and Szantyr
(: –) point out. This is what I would expect, under present assumptions, since it is the prefix
together with the DP that is structuring the PathP.
Latin as a satellite-framed language
these DPs cannot become passive subjects. These facts bring the cases of directional
DPs with unprefixed verbs together with the cases of directional PPs discussed in
section .., as we will see in section .., where I will provide a unified analysis.
.. APs
We have seen that s-framed languages like English or German admit an AP as
expression of the Core Schema, as shown by the following s-framed construction:
() German; Talmy (: )
Der Hund hat [den Schuh]Figure [kaputt]Core schema -[gebissen]Event+Co-event
the dog has the shoe in_pieces bite.PST.PART
‘The dog bit the shoe to pieces.’
In the present account, the resultative AP is, therefore, the manifestation of the vP
internal PathP. This is a natural consequence of assuming Mateu’s () reduction
of the argument structure of adjectives to that of adpositions (and recall that in Hale
and Keyser’s theory the A and P lexical heads display different projecting properties
and head different argument structures; see sections .. and ..):
() Analysis of ()
VoiceP
Voice vP
v PathP
v beiss den Schuh Path’
Path PlaceP
den Schuh Place’
Place kaputt
In Latin this option does not seem to be available, at least for s-framed constructions,
where the verb is independently associated with a Co-event root. Thus, for instance,
an example such as the following, with vacuum ‘empty’ being interpreted as the Core
Schema and with a v bundled together with the root BIB ‘drink’, is not found in this
language:
() Latin; Acedo-Matellán (: )
*Poculum vacuum bibere.
goblet.ACC.SG empty.ACC.SG drink.INF
‘To drink the globlet empty’.
The surface shape of PathP in Latin
In Chapter I will provide empirical evidence that the made-up example above
reflects a general fact of Latin—and of other similar languages like the Slavic
languages. I will also attempt an explanation of the lack of s-framed constructions
based on APs in these languages in terms of the morphological properties of v and
Path and of the adjective (section ..). Finally, I will show that APs can be part of
the PathP, as adjuncts to PlaceP, in predicates involving a non-complex change-of-
state event, that is, an event with no co-event associated:
() Latin; Plaut. Capt.
Eam [servitutem] lenem [ . . . ] reddere.
that.ACC.F.SG serfdom(F)ACC.SG mild.ACC.F.SG render.INF
‘To make that serfdom mild’.
In () the AP lenem ‘mild’ codifies the Core Schema, in that the Figure DP
servitutem ‘serfdom’ is entailed to end up in the state described by lenem. See section
.. for an analysis of these constructions.
8
See below for the question whether the dative does also express a Ground too. On the other hand,
there is a case of apparently prefix-governed accusative that is not directional:
(i) Latin; Caes. Gall. , , , in Bortolussi (: )
Flumen [ . . . ] exercitum tra-ducere maturavit
river.ACC army.ACC over-lead hasten.PRF.SG
‘He hastened to lead the army to the other side of the river.’
Predicates such as the one in (i) show two accusative DPs, one interpreted as a Figure (exercitum ‘army’),
and the other interpreted as the Ground of the preverb (flumen ‘river’). Crucially, this latter accusative is
not a directional accusative like those discussed in section ... See Lehmann (), Miller (),
Acedo-Matellán (), Bortolussi (), among others, for a description of traduco-predicates and
Acedo-Matellán () for a DM analysis.
Latin as a satellite-framed language
9
Caha () has proposed an analysis of this accusative/oblique alternation, and an analogous one in
Dutch, based exclusively on the internal structure of the PP, which hosts two different heads providing
either accusative or oblique case. In this sense, the empirical fact is missed that accusative case in
directional PPs headed by simple locative prepositions in German correlates with the fact that the PP is
merged VP-internally (Gehrke : ff.).
Latin as a satellite-framed language
However, as Ernout and Thomas (: ) later point out, there are examples where
the correlations human/dative and non-human/PP do not hold. Thus, in () a
human goal is expressed as a PP headed by ad ‘at’ and in () a non-human goal is
expressed as a dative DP in a predicate involving the prefix in- ‘in’:
() Latin; Plaut. Epid.
Illum [ . . . ] ad-ducam huc ad te.
him.ACC at-lead.FUT.SG to_here at you.ACC
‘I will bring him to you here.’
() Latin; Caes. Gall. , ,
Aggeri ignem in-ferebant.
rampart.DAT fire.ACC in-carry.IPFV.PL
‘They were carrying fire to the rampart.’
In particular, as regards the prose example in (), it is difficult to maintain the view
that the dative expresses ‘interest’, as interpreted by Rubio Fernández and González
Rolán (: ) for the following poetic example:
() Latin; Verg. Aen. ,
Silici scintillam ex-cudit.
stone.DAT spark.ACC out-beat.SG
‘He beat a spark out of the stone.’
Lehmann (: ff.) claims that the directional dative is somehow governed by
the prefix, and that the dative case is assigned to avoid the coexistence, in transitive
predicates, of two accusatives: one corresponding to the object and one introduced by
the prefix. This ‘double accusative filter’ is theoretically dubious and seems to be also
empirically incorrect, in that it fails to predict the use of the dative in cases in which
no double-accusative scenario emerges: motion predicates without a prefix (see ())
and transitive predicates with an ablative-governing prefix—cf. () above, and see
Ernout and Thomas (: –) for more examples.
Here I propose an analysis of the dative DP being merged above the projection
where directionality is encoded (PathP) but at the same time capturing the fact that it
is somehow interpreted as being governed by the prefix—as the Ground of the
motion event. My analysis is inspired by that proposed by Oya () for some
constructions in German involving a goal dative and the particle zu ‘to’:
() German; Oya (: )
Ich warf dem Kind den Ball zu.
I threw the.DAT child the.ACC ball to
‘I threw the ball to the child.’
In this sentence the dative dem Kind ‘the child’ is interpreted as the Ground of the
preposition/particle zu ‘to’. Oya (), following Olsen (), points out that
Latin as a satellite-framed language
although this particle assigns dative in German, the directional dative does not
originate as its complement, and that it is in fact higher than the accusative (cf.
den Ball ‘the ball’), as in transfer of possession predicates headed by verbs like geben
‘give’ or schenken ‘give as a gift’. Following this line of thought, Oya () hypothe-
sizes that the dative is interpreted as inalienably possessing a null nominal merged as
the complement of the particle zu ‘to’ and referring to someone’s body. The inter-
pretation of the dative as the Ground of the preposition is nothing more than an
effect of the possessive relation between the dative and this null nominal. I think that
Oya’s analysis can be made more general. In particular, datives in motion construc-
tions can be modelled as inalienably possessing a particular region of the goal object,
by virtue of which they come to be understood as the goal object itself. Interestingly,
languages that allow datives to express possession also feature a dative used in
combination with locative expressions specifying the region of a Ground object:
() German
a. Gestern hatte ein Barbier ihr das Haar geschnitten.
yesterday had a.NOM barber DAT.F.SG the.ACC hair cut
‘Yesterday a barber had cut her hair.’
b. Die schwere Tür ist ihm hin-auf-gefallen.
the.NOM heavy.NOM door is DAT.M.SG hither-on-fallen
‘The heavy door has fallen on top of him.’
In Latin we also find this correlation, the possessive dative being well established in
the literature (cf., for instance, Fay ; Löfstedt ). In the next example the
reference of the dative Ascanio ‘Ascanius’ is actually understood as both the possessor
of the PP per membra ‘through the limbs’ and as the Ground of in-rigat ‘in-pours’:
() Latin; Verg. Aen. ,
Venus Ascanio placidam per membra quietem
Venus.NOM Ascanius.DAT placid.ACC.F.SG through limbs.A rest(F).ACC.SG
in-rigat
in-pour.SG
‘Venus pours a placid tranquillity through Ascanius’ limbs.’
This example suggests both the possessive nature of the dative used with a prefixed verb
and also that it must be merged higher than Path, provided that an accusative-marked
PP (per membra ‘through the limbs’), by hypothesis an adjunct to PathP, is interpreted
under its scope. In order to implement the analysis I assume that the dative is
introduced as the specifier of an applicative head (Cuervo ; Pylkkänen )
merged above PathP but under v:10
10
This analysis echoes that proposed by Cuervo (: ) for so-called Affected Applicatives, in
which ‘[t]he dative DP is applied to the end state of the object DP [ . . . ]. The dative DP is the “possessor” of
The surface shape of PathP in Latin
Venus Voice’
Voice vP
v ApplP
v rig Ascanio Appl’
Appl PathP
pP PathP
Place in
For the cases in which the directional dative is used with an unprefixed verb, I assume
that it takes scope too over the final location represented by the verbal root merged as
Compl-Place:
() Analysis of ()
vP
v ApplP
caelo Appl’
Appl PathP
clamor Path’
Path PlaceP
clamor Place’
Place i
the end state of the object.’ From a localistic point of view, as advocated here, the ‘end state of the object’
(i.e., the Figure) is equivalent to its final location.
Latin as a satellite-framed language
The analysis of the directional dative presented in this section explains the directional
interpretation of the dative while avoiding the problems of both Lehmann’s ()
account, on the one hand, and Ernout and Thomas’s () and Rubio Fernández
and González Rolán’s (), on the other. Thus, the dative is neither governed by
the preverb, in any sense, nor is it a benefactive/malefactive dative, since it is merged
below v, and is therefore not understood as being related with the event externally
(see Pylkkänen ). See section ..., for an extension of this analysis to the
dative of transfer predicates like do ‘give’.
11
Beavers, Levin, and Tham (: ) claim that languages acknowledged as v-framed, such as
Spanish or Japanese, do allow CDMCs with bounded paths, specifically by using elements meaning ‘until’
or ‘up to’, like Spanish hasta, to convey a bounded path and successfully combine with a manner-of-
motion verb. For space reasons I cannot refute their arguments here. See Narasimhan, Di Tomaso, and
Verspoor () and Real Puigdollers () for relevant discussion. See Inagaki () for an early
proposal of Japanese made ‘until’ as encoding final location.
Latin as a satellite-framed language
12
See Zubizarreta and Oh (: ff.) for more related evidence from Korean.
S-framed constructions in Latin
of Talmy’s typology as correct, and I will take the difference between bounded and
unbounded directional expressions to be configurational in nature. Since my aim in
this section is to show the relevance of CDMCs in characterizing Latin as s-framed,
I will restrict that name to constructions involving a bounded directional element.13
13
See Folli and Harley () for the view that (transitive) CDMCs do not necessarily involve telicity.
Latin as a satellite-framed language
All of the above examples involve a predicate the telicity of which is made evident by
the licensing of a specific expression: subito ‘suddenly’, simul ‘at once’, simulatque, ‘as
soon as’, and repente ‘suddenly’. These adverbials are not possible in predicates
expressing a simple activity, which, on the other hand, license durative adverbials
such as per-phrases with a time measure expression, corresponding to English for-
adverbials, as shown in the following examples through per aliquot dies ‘for some
days’ and diu ‘for a long time’:
() Latin; Plin. Nat. ,
Per aliquot dies vagari.
for some days.ACC wander.INF
‘That it wanders for some days’.
() Latin; Ov. Am. , ,
Diu lacrimae fluxere per ora.
For_long tears.NOM flow.PRF.PL through face.ACC
‘Tears flowed down her face for a long time.’
I assume that the difference between examples such as those in () to () and
examples such as those in () and () is configurationally represented. In par-
ticular, I claim that CDMCs are unaccusative predicates. I illustrate this with an
analysis of ():
() Analysis of ()
[vP [v v CURR] [PathP ipse [Path’ Path [PlaceP ipse [Place’ Place AD]]]]]
The subject of the construction originates as a Figure in Spec-Place. Here it enters
into a predicative relation with the root AD, which refers to a place coreferent with
one already present in the discourse (as is also understood in the English rendition
‘He arrives in haste’). The entailment that the Figure effectively ends up in the
location encoded by PlaceP is licensed by the fact that the predicate incorporates a
PathP projection, which introduces a transition in the event. In turn, the quantity DP
ipse ‘he himself ’ rises to Spec-Path and is interpreted as a Measurer of that transition,
which is not over until ipse ‘he himself ’ is at the location referred to by AD. Telicity is
licensed thereby, as evidenced by the adverbial subito ‘suddenly’. Since no Voice head
is projected, ipse ‘he himself ’ is not assigned accusative case, and raises to T, where it
is assigned nominative case. Finally, the root of the verb is here an adjunct to the
eventive head v, and is interpreted, as such, as a Manner Co-event. The English
translation provided faithfully reflects this fact, since the celerity of the motion event
is expressed there as an adjunct (in haste).14 In turn, the predicates in () and (),
14
Cf. also Serbat’s (a: ) proposal that a prefixed verb like re-gredior, literally “back-walk” should
be glossed as ‘revenir en marchant’, that is ‘go back while walking’, rather than as ‘aller vers l’arrière’, that is,
‘go towards the back’.
S-framed constructions in Latin
which express activities, rather than accomplishments, are claimed to have the
following unergative structure:
() Analysis of ()
[VoiceP pro [Voice’ Voice [vP v NAVIG]]]
Here the subject (a pro, in this case) is not a Figure, but an Originator, since it
originates at Spec-Voice. The root of the verb is not adjunct to v, but a complement,
and is interpreted as an Effected Object.
I shall not attempt an analysis of the results obtained. In particular, I shall not
provide an account of the relation between unaccusativity and unergativity and the
relevant tests. My only (modest) aim is to show that two unaccusativity/unergativity
diagnostics which have been applied in other languages—the cognate object and
measure phrase diagnostics and the agent noun diagnostics—also work for Latin.
First, I show how CDMCs do not allow a certain class of ‘objects’ that have been
independently shown to be allowed only with unergative predicates: cognate objects
and measure phrases.
Cognate objects, or internal objects in the Latin linguistics tradition (Hofmann and
Szantyr : ; Bortolussi ; Pinkster : ; Serbat b), are objects that
share the same root as the verb with which they appear. For instance, in the predicate
of the following example the accusative object vitam ‘life’ shares the same root as the
verb vivo ‘live’:
() Latin; Ter. Ad.
Vitam duram [ . . . ] vixi.
life.ACC hard.ACC live.PRF.SG
‘I have lived a hard life.’
Several authors (Larson ; Massam ; Levin and Rappaport Hovav ) have
proposed that cognate objects are only allowed with unergative verbs. Unaccusative
verbs do not license them, as shown in the following examples:15
() Levin and Rappaport Hovav (: , )
a. *The glass broke a crooked break.
b. *The apples fell a smooth fall.
c. *She arrived a glamorous arrival.
Measure phrases (the so-called accusative of extension; cf., for example, Ernout and
Thomas : ) are quantified NPs that behave, partly, as standard objects.
Importantly, as pointed out by Real Puigdollers (), measure phrases also resist
appearing in unaccusative predicates (see ()a), but are perfectly normal in
unergative ones (see ()b):
() Catalan; Real Puigdollers (: )
a. *El Pere arriba tres metres del seu poble.
the Pere arrives three metres from=the his village
b. El Josep camina quatre quilòmetres.
the Josep walks four kilometres
‘Josep walks four kilometres.’
15
See Kuno and Takami (: ff.) for a different view.
S-framed constructions in Latin
16
An anonymous reviewer points out that this test has counterexamples such as early arriver, based on
unaccusative arrive. Crucially, however, the expression is felicitous to the extent that an adverb (early) is
used. I hypothesize that the structure involved in the -er nominalization imposes an agentive interpretation
with which arrive early (intentionally), but not arrive, is compatible.
S-framed constructions in Latin
17
Interestingly, an unaccusativity test standard in Romance languages, namely, the licensing of past
participles in absolute constructions (see Burzio for Italian, Legendre for French, and
Mendikoetxea for Spanish), does not seem to pick out the class of unaccusatives in Latin, but, rather,
that of intransitive deponent verbs like morior ‘die’ (in addition, of course, to that of transitive non-
deponent verbs like mitto ‘send’). Thus, as far as I have been able to check for the verbs in () and (),
only the deponent labor ‘slip, slide, fall’ and its prefixed variants allow participles used as adjectives. See
Gianollo (: ff.) for the same observation. See, for constructions involving absolute participles in
Latin, Bolkestein (, ); Lavency (); and Coleman (), among others.
Latin as a satellite-framed language
UOCs, in particular those involving objects, have been analysed by Mateu (a) as
s-framed constructions, after attesting the fact that they are not allowed in v-framed
languages. I illustrate with the anomalous Catalan renditions of the predicates above:
() Catalan renditions of ()
a. *La Sue cridà en John sord.
b. *La Sue lluità en John a terra.
c. *La Sue treballà el seu deute fora.
() Catalan renditions of ()
a. *El meu abric (es) plou humit.
b. *El paperet (es) bufa al forat.
c. *La planta (es) neva a sota.
In a nutshell, Mateu (b) adopts a Hoekstrian (Hoekstra ) analysis in terms
of a Small Clause. UOCs, then, involve an abstract causative V and a Small Clause
complement whose subject is the unselected object of the UOC and whose predicate
is the piece of the UOC licensing the unselected object: a particle, a PP, or an
AP. The Small Clause is headed by a prepositional head. Since in s-framed languages
(like English and Dutch above) this prepositional head is realized independently
from the eventive V head, V may host an independent unergative structure codifying
the accompanying Co-event (a shouting event, for instance, in ()a). On the other
hand, v-framed languages, like Catalan, do not license the constructions, since the
prepositional head is conflated into V and conflation of an independent element is
incompatible with this circumstance. I assume a similar analysis:
() Analysis of ()a
[VoiceP Sue [Voice’ Voice [vP [v v SHOUT] [PathP John [Path’ Path [PlaceP John [Place’
Place DEAF]]]]]]]
Motivation for this analysis is mainly based on the semantic interpretation of the
predicative piece that licenses the unselected object. In (), for instance, the
inference is licensed that as the result of some event originated by Sue, which is
identified with a shouting event (see the adjunction relation of SHOUT with v), John
ends up deaf. John, interpreted as Figure in the predicative relation structured around
Place, is, in turn, interpreted as a Measurer of the event in Spec-Path position.
In this book I will analyse two types of UOCs in Latin: Figure UOCs (this section)
and Ground UOCs (section ..). In the former type the internal argument corres-
ponds to the Figure, that is, to the DP merged as Spec-Place. In the latter type the
internal argument corresponds to (and is interpreted as) the Ground, that is, the DP
merged as Compl-Place. What unifies both types is that, whether Figure or Ground,
this DP is internally merged as Spec-Path, where it is interpreted as a Measurer.
Different constructions will be shown to be UOCs in confronting them with
S-framed constructions in Latin
constructions involving the same verb but in the absence of a special context. The set
of semantic and syntactic differences between both types of constructions will be
established and shown to be naturally derived from the status of UOCs as change
predicates involving a PathP. Additionally, UOCs are presented as an optimal case
study to show how a neo-constructionist view of argument structure naturally
predicts that the licensing conditions in predicates—in the current case, the licensing
of objects—depend on the syntactically assembled pieces they are made of, and not
on a single projecting nucleus (the verb).
As just mentioned, Figure UOCs feature an internal argument interpreted as
Figure. Figure UOCs are very commonly represented in Latin in the form of predi-
cates headed by a prefixed verb and accompanied, sometimes, by a directional DP or
PP. I will be illustrating Figure UOCs through the prefixes ex-, ab-, and in-, and I will
show the semantic and syntactic differences between the unprefixed and prefixed
predicates. Then I will focus on a series of particular properties of these Figure UOCs:
the licensing of null objects, case and situation-aspect properties, and scopal
relations between prefix and verb. I shall argue that these properties naturally derive
from a syntactic neo-constructionist perspective on argument structure and word
formation.
... The syntax and semantics of prefixed vs unprefixed verbs In this section
I illustrate Figure UOCs in Latin through predicates headed by ex-, ab-, and in-
prefixed verbs. I show the great elasticity of verbs by pointing out the semantic and
syntactic differences between the prefixed verbs and their unprefixed counterparts.18
The prefix ex- (with the variant e-) has the core meaning of ‘out’. This is shown by
the following UOCs:
() Latin; Cato, Agr. ,
Qui oletum saepissime et altissime miscebit,
who.NOM olive-tree.ACC often.SUPERL and deeply.SUPERL mix.FUT.SG
is tenuissimas radices ex-arabit.
he.NOM slender.SUPERL.ACC.PL root.ACC.PL out-plough.FUT.SG
‘He who works his olives very often and very deep will plough out the most
slender roots.’
() Latin; Verg. Aen. ,
Immanisque columnas rupibus ex-cidunt.
huge.ACC.PL=and column.ACC.PL rock.ABL.PL out-cut.PL
‘And they hew huge columns out of rocks.’
18
For more in-depth studies of the semantics of Latin verbal prefixes, see Pottier () and García
Hernández ().
Latin as a satellite-framed language
tilling ground and in () it is the snakes’ own bodies. All the cases involve the
projection of a PathP, the adjunction of a Manner root to v, introducing the Co-
event, and root EX ‘out’, merged at Compl-Place, where it is interpreted as a Ground
(a Terminal Ground, in fact, since PlaceP is embedded under PathP, encoding a
transition), and predicates such as that of () emerge, where the specific reference
of the Ground is calculated contextually:
() Analysis of ()
[VoiceP is [Voice’ Voice [vP [v v AR] [PathP tenuissimas radices [Path’ Path [PlaceP
tenuissimas radices [Place’ Place EX]]]]]]]
The abstract final location expressed by the prefixal root may be further specified by
an adjunct, as in ():
() Analysis of ()
VoiceP
pro Voice’
Voice vP
v PathP
pP PlaceP
p rupibus
immanis columnas Place’
Place ex
My analysis of prefixed verbs in Latin can claim to formalize Serbat’s (a: ff.)
important insight that in a prefixed predicate the verb is a secondary predicate, while
the prefix is the main predicate,19 although it clashes with most of Latin linguistics
tradition, in which the prefixed verb is the result of the affixation of a prepositional
element to a pre-existing simple verb20 (Donatus, th century AD; Priscian, th century
19
I reject, however, Serbat’s (a: ) implementation, where the verbal inflectional morphology
changes the prefix into a verbal predicate, since I dissociate the notion of morphosyntactic category from
that of predicate.
20
Note, also, that my analysis of Latin prefixation is orthogonal to the question of whether it is a case of
composition or of derivation, since these are not primitive concepts in the theory of Distributed
Latin as a satellite-framed language
AD; Brachet : ; Romagno : ). The analysis also assumes that whatever
semantic relation is established between the object and the verb, it is the result of, on
the one hand, the interpretation of the structure in which they appear (the structural
semantics) and the roots merged within that structure (the encyclopaedic semantics):
there is no (direct) thematic relation between the object and the verb. Thus, the object
is always interpreted as a subject of a predication established by an abstract head,
Place. In turn, it is also interpreted in the structures above as a Measurer of the event.
In (), for instance, the event is over when the huge columns are literally out of the
rocks: there is a direct relation between the quantity expressed by columnas ‘columns’
and the quantification of the event itself. In turn, the interpretation of the verb relies
on the existence of an event introduced by v and a Manner Co-event expressed by the
root adjoined to it. Note, then, that the conceptual dimensions of the verb and of
the object are completely severed from each other. In (), for instance, there is no
direct conceptual relation between the hewing activity and the columns, nor are they
affected thereby. Although this might seem counter-intuitive at first sight, it is
supported by cases of UOCs where the simple verb, outside the UOC, does not
usually take any object. In the above examples there is actually one such case: that of
(), headed by ex-tussio “out-cough” ‘expectorate (something) through coughing’.
Simple tussio ‘cough’ is not registered to allow for any objects (Gaffiot ). The
meaning of extussio is furthermore not licensed with an independent ex-PP. The rest
of the examples constitute cases of ‘weak’ UOCs, in that their simple counterparts
can be transitive but do not license the same type of object. These UOCs, however,
are also able to refute the inference that makes objects in prefixed-verb predicates
seem to be affected by the action conveyed by the verb. Thus, in () the seeds,
necessarily, are not damaged by a rubbing or grinding action (conveyed by tero
‘grind’), as are, naturally, the husks out of which they come. The same happens with
ex-aro “out-plough” in (): the roots are not ploughed (AR), but extracted (EX) upon
ploughing (AR).
Under the present assumptions, the verbs are expected to show, on the surface, a
great elasticity, since roots may, in principle, be merged in any context (where they
are structurally admitted, that is: as complements or adjuncts of functional heads).
This elasticity is, I argue, restricted by clashes between the encyclopaedic content of a
root and the interpretation of the position it occupies in the structure. Analyses
attributing grammatical features to verbs (roots) fail to predict this elasticity. For
instance, Lehmann (: ) proposes that the argument structure of the base verb
should be kept in the prefixed counterpart, that is, added to that of the preverb—see
also Carvalho (: ). However, this hypothesis cannot explain why an obliga-
torily transitive verb like rumpo ‘break’ may be used as an intransitive when prefixed:
Morphology. For phonological arguments in favour of the former view, see Heslin (). For phono-
logical, semantic, and syntactic arguments in favour of the latter view, see Oniga ().
S-framed constructions in Latin
() yields an interpretation where the ploughing activity is exerted on terram ‘the
earth’ merged as an adjunct to vP, with no resulting state entailed (see section ...).
Certain ex-verbs exist which head predicates where the Ground, rather than a physical
entity, is someone’s spiritual dimension or their possessions. They thus imply that
something (the Figure object) is obtained from someone by some activity, specified by
the root merged as an adjunct to the eventive v head. For instance, in the following
examples things are obtained through flattery, enchantment, and caresses, respectively:
() Latin; Liv. , ,
Neque enim omnia emebat aut e-blandiebatur
nor in_fact all.ACC.N.PL buy.IPFV.SG or out-flatter.IPFV.SG
‘Nor did he acquire everything by money or flattery.’
() Latin; Sen. Nat. b, ,
Ne quis alienos fructus
lest anybody.NOM of_another.ACC.PL fruit.ACC.PL
ex-cantassit.
out-enchant.PLUPRF.SBJV.SG
‘Lest anyone should obtain someone else’s fruits through enchantment’.
Latin as a satellite-framed language
In the examples seen so far the Ground is a concrete entity. But it can also be
understood more abstractly, as a general ‘here and now’, facilitating a ‘disappearance’
sense for the prefix:
() Latin; Cic. Phil. ,
E-dormi crapulam, inquam.
out-sleep.IMP.SG intoxication.ACC say.SG
‘Sleep off the intoxication, I said.’
() Latin; Cato, Agr. ,
Usque coquito, dum dimidium ex-coquas.
until cook.IMP.FUT.SG until half.ACC out-cook.SBJV.SG
‘Boil it until you boil half of it away.’
In () there is an example of simple dormio ‘sleep’. Since it cannot take objects in
the accusative, predicates involving e-dormio “out-sleep” constitute one of those
cases of UOCs where the prefix is ostensibly facilitating the projection of an accusa-
tive object:
() Latin; Ov. Rem.
Thalamo dormimus in illo.
bridal bed(M)ABL.SG sleep.PL in that.ABL.M.SG
‘We slept in that bridal bed.’
The verb ex-coquo “out-boil”, on the other hand, already appeared in () (repeated
below as ()) as an example of UOC that, although hyperbolically used, involves a
concrete entity as Ground (someone’s fortune):
() Latin; Plaut. Capt.
HEGIO: Quid diuitiae? Sunt ne opimae?
what.ACC richness(F)NOM.PL are PART.INTER abundant.F.NOM.PL
PHILOCRATES: Vnde ex-coquat sebum senex.
whence out-boil.SBJV.SG tallow.ACC old_man.NOM
‘HEGIO: What about his riches? Are they abundant?—PHILOCRATES: So much
that the old rascal could melt out the tallow.’
The semantic difference between () and () consists, then, in the fact that in the
former the object undergoes disappearance, while in the latter it happens to appear
out of somewhere, this location being identified by the pronoun unde ‘whence’. In
both cases the root COQU ‘boil’ is merged as an adjunct modifier of the change-of-state
predicate headed by the Path head: the boiling/melting event is in both cases a
manner co-event. The difference lies, I argue, in the nature of the element identified
as the location: non-referential in () and referential in (). Specifically, a
plausible analysis involves merging the root of the prefix as Compl-Place in both
Latin as a satellite-framed language
cases, while having the relative adverb unde ‘whence’ an adjunct to PlaceP in (), as
shown below:
() Analyses of () and ()
a. [VoiceP (tu) [Voice’ Voice [vP [v v COQU] [PathP dimidium [Path’ Path [PlaceP
dimidium [Place’ Place EX]]]]]]]
b. [VoiceP (tu) [Voice’ Voice [vP [v v COQU] [PathP sebum [Path’ Path [PlaceP unde
[PlaceP sebum [Place’ Place EX]]]]]]]]
A specialization of the disappearance sense is found in verbs where the Figure goes
away through expenditure. The way that expenditure is carried out is, as is to be
expected, expressed by the verbal root:
() Latin; Hor. Sat. , ,
Filius [ . . . ] haec [ . . . ] ut e-bibat [ . . . ] custodis?
son.NOM this.ACC.PL that out-drink.SBJV.SG guard.SG
‘You guard [these possessions] to the end that thy son guzzles them all up?’
() Latin; Plaut. Trin.
LESBONICUS: Quid factumst eo [minas quadraginta]?
what.NOM made=is it.ABL mina.ACC.PL forty
Stasimus: Com-essum, ex-potum,
with-eat.PTCP.PFV.NOM.N.SG out-drink.PTCP.PFV.NOM.N.SG
ex-unctum, e-lotum
out-anoint.PTCP.PFV.NOM.N.SG out-wash.PTCP.PFV.NOM.N.SG
in balineis.
in bath.ABL.PL
‘LESBONICUS: What has been done with it (forty minas)?—STASIMUS: It has
been eaten, drunk up, spent away in unguents, washed away in baths.’
The kind of object appearing with the simple counterparts of these verbs is quite
different. I capitalize here on the differences between bibo ‘drink’ and e-bibo “out-
drink” ‘drink up’ (see Vendryès and Brachet : ) in an attempt to provide
new evidence in support of a syntactic analysis of these phenomena. The following is
an example of simple bibo ‘drink’ where there is no specific entailment that the water
is exhausted through drinking. On the contrary, bibo ‘drink’ expresses an activity:
() Latin; Cat. Agr. ,
Per aestatem boues aquam bonam
through summer.ACC cow.NOM.PL water.ACC good.ACC
et liquidam bibant semper curato.
and clear.ACC drink.SBJV.PL always care.IMP.FUT.SG
‘One must always see to it that cows drink good and clear water all through
the summer.’
S-framed constructions in Latin
The contrast is particularly dramatic in the next example. I have included the whole
paragraph, since it involves both bibo ‘drink’ and e-bibo “out-drink”:
() Latin; Petr. Sat. ,
‘Quid? ego’ inquit ‘non sum dignus qui bibam?’
what I say.SG not am worthy.NOM who.NOM drink.SBJV.SG
ancilla risu meo prodita complosit
serf.NOM.F.SG laugh.ABL my.ABL betray.PTCP.PFV.NOM.F.SG clap.PRF.SG
manus et ‘apposui quidem adulescens, solus
hand.ACC.PL and serve_up.PRF.SG certainly youth.VOC alone.NOM.M.SG
tantum medicamentum e-bibisti?’
so_much.ACC medicine.ACC out-drink.PRF.SG
‘ita ne est?’ inquit Quartilla ‘quicquid satyrii
thus PART.INTERR is say.SG Quartilla.NOM whatever.ACC satyrion.GEN
fuit, Encolpius e-bibit?’
be.PRF.SG Encolpius.NOM out-drink.PRF.SG
‘ “Well, then, why should I not deserve to drink?” The serf, betrayed by my
laugh, clapped her hands and (said) “I have served you up already, youth. By
the way, have you drunk up such an amount of medicine all by yourself?”
“Really?”, said Quartilla, “Has Encolpius drunk up all the satyrion21?” ’
Simple bibam ‘I drink’ is interpreted as an atelic activity. Here the root is merged as a
Compl-v, and is interpreted as an Effected Object:
() Analysis of simple bibo
[VoiceP qui [Voice’ v BIB]]
However, the two instances of e-bibo “out-drink” express the exhaustion of the liquid,
as reflected on the translations. I propose that they correspond to a different structure:
() Analysis of tantum medicamentum ebibisti (in ())
[VoiceP (tu) [Voice’ Voice [vP [v v BIB] [PathP tantum medicamentum [Path’ Path
[PlaceP tantum medicamentum [Place’ Place EX]]]]]]]
The prefix originates as a root merged as Compl-Place; here it is understood as a
Terminal Ground, expressing the final state of the Figure tantum medicamentum: the
state of disappearance (akin to the one encoded by up in English drink the wine up).
The DP in Spec-Place rises to Spec-Path. There it is interpreted as Measurer: when
the amount described by tantum medicamentum ‘so much medicine’ reaches the
state described by the root EX ‘out’, the event, specified as a drinking event by the
adjunct root BIB ‘drink’, is over. I shall come back to the bibo/ebibo difference in
relation to the licensing of null objects in section ....
21
An aphrodisiac drink.
Latin as a satellite-framed language
I turn now to the prefix ab- (with variants a- and abs-) ‘off, away’, which presents a
central meaning of ‘separation from a surface’. This prefix is widely used in prefix-
ation to surface-contact verbs indicating the way in which the separation takes place:
() Latin; Tac. Hist. , ,
Is balineas ab-luendo
he.NOM bath.ACC.PL off-wash.PTCP.FUT.PASS.DAT.M.SG
cruori propere petit.22
blood(M)DAT.SG hastily head.SG
‘He hastened to the baths to wash off the blood.’
() Latin; Colum. Arb.
Sarmenta [ . . . ] arida [ . . . ] dolabra ab-radito.
shoot.ACC.PL dry.ACC.PL hatchet.ABL off-razor.IMP.FUT.SG
‘The dry vine shoots are to be razored off with a hatchet.’
() Latin; Liv. , ,
Inspectum vulnus
examine.PTCP.PFV.NOM.N.SG wound(N)NOM.SG
abs-terso cruore.
off-wipe.PTCP.PFV.ABL.M.SG blood(M)ABL.SG
‘That the wound had been examined after wiping the blood off ’.
() Latin; Hor. Sat. , ,
Cara piscis a-verrere mensa.
expensive.ABL.F.SG fish.ACC.PL off-sweep.INF stand(F)ABL.SG
‘To sweep away the fish from an expensive stand’.
The Ground in the above examples, corresponds to a surface whose identity is
discursively retrieved in (), (), and () (coreferent with vulnus ‘wound’);
in () the prefix coexists with an overtly expressed Ground in the ablative (cara
[ . . . ] mensa ‘an expensive stand’).
As is to be expected, the roots we find in the prefixed verbs above may appear in
other syntactic environments. In the following examples, tergo ‘wipe’ and verro ‘sweep’
appear in an unergative form, without any object, and with an activity interpretation:
() Latin; Cic. Parad. ,
Qui tergent, qui ungunt, qui verrunt.
who.NOM.PL wipe.PL who.NOM.PL anoint.PL who.NOM.PL sweep.PL
‘Those who wipe, those who anoint, those who sweep.’
The roots can appear in transitive predicates headed by simple verbs:
22
Abluendo cruori ‘to wash off the blood’ is a so-called gerundive construction, with a passive verbal
adjective abluendo agreeing with dative cruori, which is the logical object of the construction.
S-framed constructions in Latin
() Latin; transitive uses of some of the simple verbs in () to ()
a. Cic. Leg. ,
Mulieres genas ne radunto.
woman.NOM.PL cheek.ACC.PL not razor.IMP.FUT.PL
‘Do not let the women scratch their cheeks.’
b. Verg. Aen. ,
Clipeos [ . . . ] tergent arvina pingui.
shield.ACC.PL wipe.PL grease.ABL thick.ABL
‘They polish the shields with thick grease.’
c. Plaut. Merc.
Nil opust nobis ancilla,
nothing.NOM is_needed us.DAT slave_girl.NOM
nisi quae [ . . . ] aedis uerrat.
except who.NOM.F.SG house.ACC.PL sweep.SBJV.SG
‘We need nothing but a slave girl who can sweep the house.’
While the objects in predicates headed by prefixed verbs are understood as entities
which, through different process, become separated from a surface (explicit or not),
the ones in ()a to ()c refer, on the contrary, to surfaces on which the action
portrayed by the verb is exerted. Note, for instance, that genas ‘cheeks’, in ()a, are
not cut off from anywhere, as is the case with sarmenta ‘vine shoots’ in (). Instead,
genas ‘cheeks’ in ()a are understood as surfaces where a scratching action
takes place.
As is the case with ex- ‘out’, the sense of ab- as ‘separation from a surface’ meaning
is easily extended to a disappearance meaning, including the ‘spend by X-ing’ sense
we saw before. In this case, the Ground is understood deictically, as a general ‘here
and now’:
() Latin; Apul. Met. ,
Iucundiora [ . . . ] ab-ligurribam dulcia.
delicious.COMPAR.ACC.N.PL away-lick.IPFV.SG sweet.ACC.PL
‘I used to lick away rather delicious sweets.’
() Latin; Cat. Agr. ,
Omne caseum cum melle ab-usus eris.
whole.ACC cheese.ACC with honey.ABL away-use.FUT.SG
‘You will have used up all the cheese with honey.’
() Latin; Ter. Eun.
Patria qui ab-ligurrierat bona.
paternal.ACC.N.PL who.NOM away-lick.PLUPRF.SG good(N)ACC.PL
‘Who had wasted the paternal goods luxuriously.’
Latin as a satellite-framed language
23
Silphium, -ii: a plant. Cato is describing a recipe for cabbage, into which silphium must be grated.
S-framed constructions in Latin
The objects in the above examples are quite evidently semantically unselected by the
base verbs. For instance, in () a vine cannot be ploughed, but introduced
somewhere by ploughing. Likewise, in () acta ‘acts’ cannot be caesa ‘cut’, but
can be in-cisa “in-cut” ‘engraved’. In () the plant silphium is not the Patient of
a scraping event, rather it is a Figure which changes location through scraping.
Similarly in () notae ‘motifs, designs’ cannot be woven, but they can be woven
into the fabric, that is, introduced into the fabric by weaving. The unprefixed
counterparts of these verbs show completely different semantic relations with their
objects (see (); the first two examples are passives) and some of them are found in
unergative environments (see ()):
() Latin; transitive uses of the simple counterparts of some of the verbs in () to ()
a. Cato, Agr. ,
[Posse] hortum fodiri.
can.INF yard.NOM dig.INF.PASS
‘The garden may be dug.’
b. Varr. Rust. , ,
Terra [ . . . ] facile frietur.
earth.NOM easily grind.SBJV.PASS.SG
‘Earth crumbles easily.’
c. Ter. Haut.
Texentem telam studiose ipsam offendimus.
weave.PTCP.PRS.ACC.F.SG cloth.ACC painstakingly her.ACC find.PRF.PL
‘We found her painstakingly weaving a cloth.’
() Latin; unergative use of the unprefixed counterparts of two of the verbs in
() to (), Ter. Haut.
Te in fundo conspicer fodere aut arare.
you.ACC in farm.ABL spot.INF dig.INF or plough.INF
‘(I see) you digging or ploughing on your farm.’
I make a final observation on im-misceo “in-mix”, in (). This case is interesting
because one of the usual arguments of simple misceo ‘mix’ is missing, namely, that
referring to the substance or set of things with which the object is mixed, which may
appear in the dative, ablative, or as a PP (see, respectively, ()a, ()b, and ()c);
alternatively, misceo ‘mix’ may appear with two coordinated DPs referring to the
substance being mixed together (see ()d):
() Latin; simple misceo ‘mix’
a. Ov. Met. ,
Fletumque cruori miscuit.
tear.ACC=and blood.DAT mix.PRF.SG
‘She mixed her tears with his blood.’
Latin as a satellite-framed language
b. Hor. Sat. , ,
Surrentina [ . . . ] miscet faece Falerna vina.
Surrentine.ACC.N.PL mix.SG dregs.ABL Falernian.ABL wine(N)ACC.PL
‘He mixes Surrentine wines with Falernian dregs.’
c. Cato, Agr.
Caseum cum alica [ . . . ] misceto.
cheese.ACC with spelt.ABL mix.IMP.FUT.SG
‘Mix the cheese with spelt.’
d. Plin. Nat. , ,
Vinum et aquam miscent.
wine.ACC and water.ACC mix.PL
‘They mix wine and water together.’
The syntactic environment in which im-misceo “in-mix” is found is different, and
highly predictable: it is the syntactic environment of any UOC. It features, on the one
hand, a DP, soporiferum venenum ‘soporific poison’, interpreted as Figure and as a
Measurer of the event, since the quantity of poison determines the temporal span of
the mixing event. On the other hand, a directional dative DP expresses the final point
of a spatial transition: illis cantharis ‘those jars’. The main event, then, is a transition
whereby the poison (venenum soporiferum) ends up in the jars (illis cantharis)
through a mixing event (encoded in the root MISC ‘mix’). As I discussed in section
.., I assume that the directional dative is not a Ground, syntactically, but the
specifier of an applicative head interpreted as inalienably posessing the referent of the
Ground, in this case the root IN ‘in’:
() Analysis of ()
VoiceP
pro Voice’
Voice vP
v ApplP
v misc illis cantharis Appl’
Appl PathP
Path PlaceP
Place in
S-framed constructions in Latin
Importantly, the substance, with which the venenum soporiferum ‘soporific poison’ is
mixed and which is presumably contained in the jars is not expressed; in fact, it cannot be
expressed in (), at least not as a part of the argument structure configuration in ().
This discussion shows that whatever event participants roots require as part of their
idiosyncratic content (in this case, the ‘second’ substance in a mixing event) can and must
be overriden if the structure demands it. The syntactic configuration, therefore, imposes
a certain interpretation on the root: while the unprefixed verb may identify a final state
(see ()), the prefixed counterpart must be interpreted as a co-event (see ()).24
... Conditions on the licensing of null objects: bibo ‘drink’ vs ebibo ‘drink up’ In
this section I point out a crucial syntactic difference between bibo ‘drink’ and ebibo
“out-drink” ‘drink up’, which, within the present account, receives a natural explan-
ation. Specifically, bibo ‘drink’ may appear without an object, focusing merely on a
process (often of drinking wine), as has been shown above in () and is further
shown in () (in the usage referred to in traditional grammars as absolute—cf.
Ernout and Thomas : ff.):
() Latin; Object-less bibo ‘drink’
Andr. Commoediarum fragmenta in aliis scriptis servata,
Edi bibi lusi.
eat.PRF.SG drink.PRF.SG play.PRF.SG
‘I ate, I drank, I played.’
On the contrary, the rare object-less instances of e-bibo “out-drink” found in the
corpus and shown in () appear after an entity is introduced in the discourse that
provides the reference of the object. An objective null category (represented by ei in
the examples) corresponds to the object of e-bibo “out-drink” in these instances:25
() Latin; e-bibo “out-drink” with null objects
a. Plaut. Curc.
Propino [magnum poculum]i:
bring_forth.SG big.ACC goblet.ACC
ille ei e-bibit, caput de-ponit, con-dormiscit.
he.NOM out-drink.SG head.ACC downward-put.SG together-sleep.SG
‘I bring forth a big cup to him: he gulps it empty, lays his head down, and
falls asleep.’
24
See Zeller (b) and McIntyre () for data and accounts of how Germanic particles—which
behave similarly to Latin prefixes—can neutralize the usual argument structure displayed by a verb and
oblige its internal argument to be demoted as an adjunct. I will return to this ‘demotion of arguments’ in
sections ... and ...
25
I note that ()a is not a Figure UOC. It is, rather, a Ground UOC, since the object (coindexed with
magnum poculum ‘big cup’) is interpreted as a Ground (the container out of which the wine is drunk).
However, the Figure/Ground UOC distinction is not crucial for the current purpose, namely showing the
syntactic differences between simple and prefixed verbs. I will deal with Ground UOCs in section ...
Latin as a satellite-framed language
26
Hemina, -ae: a measure of wine.
27
Discussions on null objects in Latin include Panhius (); Mulder (); Wurff (); and
Sznajder ().
S-framed constructions in Latin
... Case alternations, situation aspect, and the merging of roots I focus now on
the way prefixation changes the case-assigning properties of the predicate, and how
that change is related to the inner-aspectual interpretation of the predicate. I take the
utor/abutor ‘use’/ “away-use” ‘use up’ contrast (see example ()) as a case study.
Importantly, while ab-utor “away-use” licenses an accusative in (), repeated here
as (), the ‘object’ of utor ‘use’ appears in the ablative (see ()):
() Latin; Cat. Agr. ,
Omne caseum cum melle ab-usus eris
whole.ACC cheese.ACC with honey.ABL away-use.FUT.SG
‘You will have used up all the cheese with honey.’
() Latin; Caes. Gall. , ,
Minus idoneis equis utebantur.
less suitable.ABL.M.PL horse(M)ABL.PL use.IPFV.PL
‘They were using less suitable horses.’
In fact, my prediction is that the object of ab-utor “away-use” should appear always
in the accusative in UOCs: it sits at Spec-Path, as evidenced by the fact that it behaves
as a Measurer. In (), for instance, the event is over only when the whole amount of
cheese is used up. As a matter of fact, ab-utor “away-use” does sometimes take the
ablative case in Classical Latin. The next example, for instance, involves ablative
sagacitate ‘sagacity’ instead of accusative sagacitatem:
() Latin; Cic. Nat. deor. ,
Sagacitate canum ad utilitatem nostram ab-utimur.
sagacity.ABL dog.GEN.PL at benefit.ACC our.ACC away-use.PL
‘We (abusively) use the sagacity of dogs to our benefit.’
However, a look at Gaffiot’s () entry for abutor “away-use” reveals a possible
explanation for this double case-selection. The first sense in the entry, the only
transitive one, reads ‘use until the object disappears’. Gaffiot furthermore marks it
as archaic, providing examples from Cato, Plautus, Terence, and Sallust. This is the
sense illustrated in (). The second sense is intransitive, taking the ablative, and is a
more modern one. The definition here reads differently, however: ‘use fully, freely’ or
‘make a deviant use of something’. This is the usage relevant in (). Observe, in
addition, that the ablative, as expected, does not license a Measurer interpretation for
sagacitate ‘sagacity’ in (); in fact, as the famous Ciceronian sentence of ()
indicates, this sense of abutor as ‘make an improper use of, abuse’ is atelic, since it
licenses the durative adverbial quo usque ‘until when’:28
28
In light of these facts, claims such as Oniga’s (: ) that transitivization through prefixation of
ablative-selecting prepositions—like ab ‘away’—must be due to a phonological or semantic analogy with
verbs which are prefixed with accusative-selecting prepositions cannot be on the right track. In particular,
Latin as a satellite-framed language
he cites the case of transitive abutor as an analogy of ad-sumo “at-take” ‘take to oneself ’ (with accusative-
selecting ad ‘at’). This would imply a greater antiquity for the intransitive abutor, contrary to what Gaffiot
() documents.
29
See Zhang () for an application of this idea to the analysis of compounds in Chinese.
30
Wurmbrand () advocates a complex predicate approach for idiomatic particle-verb combinations
in German, while reserving a Small-Clause approach for cases of transparent particle-verb combinations.
I am sympathetic to her analysis, but I do not think that idiomaticity (here the possibility of retrieving
particular meanings for roots within the phase) is restricted to direct association of roots. See McIntyre ()
for a critique of Wurmbrand’s () dychotomic approach to particle-verbs.
S-framed constructions in Latin
... Scopal relations between prefix and verb Scopal effects have traditionally
been dealt with at the sentence level in discussions of configurationality. However,
within an account, such as the present one, where words are created by the syntax, we
expect there to be scopal effects within the word. I will now show that there is a group
of ab-prefixed Figure UOCs in Latin that show scopal effects affecting the prefix and
the verb. Importantly, these effects follow naturally from an account of UOCs where
the prefix is c-commanded by the v head and the root is merged as an adjunct to v.
I refer to a group of ab-verbs where the base is a communication verb and the prefix
is interpreted as a negation of sorts. I call them ab-verbs of denial:
() Latin; Pacuv. Trag.
[Eam] consanguineam esse ab-dicant.
her.ACC consanguineous.ACC be.INF away-proclaim.PL
‘They proclaim her not to share the same blood.’
() Latin; Cic. Div. ,
Cumque in quattuor partis vineam
since=and in four part.ACC.PL vine.ACC
divisisset trisque partis aves
divide.PLUPRF.SBJV.SG three.ACC=and part.ACC.PL bird.NOM.PL
ab-dixissent, quarta parte [ . . . ] mirabili
away-say.PLUPERF.SBJV.PL fourth.ABL part.ABL admirable.ABL
magnitudine uvam [ . . . ] invenit.
size.ABL grape.ACC find.PRF.SG
‘And after he had divided the vine into four parts and the birds had refused
[lit. ‘had said away’] three of them, in the fourth part he found a grape of
admirable size.’
() Latin; Plaut. Rud.
In iure ab-iurant pecuniam.
in court.ABL away-swear.PL money.ACC
‘In court they deny by oath that they have debts.’
() Latin; Plaut. Capt.
‘Ubi cenamus una?’ inquam: atque illi ab-nuont.
where sup.PL together say.SG and they away-nod.PL
‘I say, “Where shall we sup together?” And they refuse with a nod.’
These verbs involve the negation of the proposition expressed by the object (which
may take the shape of a whole proposition, as in the Exceptional Case Marking
construction of () (with accusative eam ‘her’ as the subject of the embedded
infinitive esse ‘be’) or the elided object proposition of (), or a propositionally
interpreted DP, as in () and ()). In the examples above, the base verb is,
Latin as a satellite-framed language
respectively, dico (infinitive dicare) ‘proclaim, declare’, dico (infinitive dicere) ‘say’,
iuro ‘swear’, and nuo ‘nod’. Take the case of ab-iuro “away-swear” ‘deny by oath’, in
(). Crucially, as García Hernández (: ) early observed, the negation is
understood as having narrow scope with respect to the swearing event introduced by
the root IUR, and not the other way around. Thus, () does not imply that they do
not swear that they have debts. The scopal properties of these verbs come for free in a
syntactic model, if we assume that the negation component alluded to above is
nothing but an inference from the general meaning of the prefix ab ‘away’: the v
introducing the event and being associated with the root IUR ‘swear’ is above the
PathP including the object of the predicate and the prefix. Importantly, the root of
the prefix is c-commanded by the v head introducing an event, and is predicted,
correctly, to show narrow scope with respect to that v head:
() Analysis of ()
[VoiceP pro [Voice’ Voice [vP [v v IUR] [PathP pecuniam [Path’ Path [PlaceP
pecuniam [Place’ Place AB]]]]]]]
Scope: v > ab, *ab > v
An approach conceding a preponderant role to configurationality and separating the
encyclopaedic from the structural meaning of expressions derives both the denial
interpretation of these verbs and the precise scopal effects straightforwardly.
31
Studies on these constructions in other languages include Svenonius (: ff.) on Scandinavian,
Zeller (a) on German, McIntyre () on German, McIntyre (, ) on English and German,
Svenonius () on English, Svenonius () on Russian, Mateu (b) on German, Levin and Sells
() on English (calling the particles in these constructions unpredicated particles), and Oya () on
English, German, and Dutch.
S-framed constructions in Latin
whole surface of the table for the event of wiping the dust off the table to be true. This
effect in Ground UOCs is observed by McIntyre (), who notes the contrast
between Read through a book and the Ground UOC Read a book through:
reading through a book is less thorough than reading a book through. Although the former
could exhibit the bounded reading of through in the sense that the reading encompasses
the beginning and end of the book, it is compatible with skim-reading or leaving out some
sections because there is no holistic effect to ensure that the whole book is involved.
McIntyre (: )
Observe that in Reading through a book there is apparently no Figure. In fact, McIntyre
() proposes that the whole event of reading is a Figure traversing the Path
expressed by through a book (he calls this kind of construction Event Path). What is
worth noting here is that the different position of the Ground determines the above-
mentioned holistic effect or measuring-out effect. Crucially, McIntyre’s () obser-
vation can be made stronger, by setting it in terms of (a)telicity: while read through a
book may be atelic, read a book through is necessarily telic. Similar observations on the
measurer role of Grounds in these constructions are to be found in Levin and Sells
(). As the next examples show, the quantity or non-quantity status of the object
Ground is what determines, respectively, telicity and atelicity in the resultant predicate:
() Levin and Sells (: )
a. She wiped the counter off in/*?for ten minutes.
b. She wiped glass off *in/for two hours.
These facts are easily accounted for in my theory, since the Figure and Ground
interpretations of a DP are dissociated from its role in the calculation of situation
aspect. Thus, we expect either one of them to be available to be merged as Spec-Path
and to be interpreted as Measurer. I illustrate with the analysis of ():
() Analysis of ()
VoiceP
Sue Voice’
Voice vP
v PathP
Place out
S-framed constructions in Latin
The DP the bucket is originally merged as Compl-Place, and is, therefore, interpreted
as a Ground. Path raises the nearest DP in its c-command domain to its specifier (see
section ...). This DP is usually the one sitting at Spec-Place, the Figure, by a
minimality condition, but when the Figure is missing, there is no DP available other
than the Ground. It is at Spec-Path that the bucket is interpreted as a Measurer. As for
case, the bucket receives the same treatment as any other DP at Spec-Path: it gets
accusative case if Voice is projected, as in this instance. However, we will see in
section ... that in unaccusative Ground UOCs, with no Voice head projected, the
Ground ends up receiving nominative case, as one would expect.
Note that a case-account, such as that of Svenonius (), is unable to explain
why the Ground is interpreted as a Measurer only when it appears as the object of the
verb. Indeed, this Measurer interpretation cannot be attributed to the accusative case
itself, since there are accusative-marked DPs that are not interpreted as Measurers
(such as Peter in the next sentences):
() Non-measuring accusatives
a. John loved Peter (for years).
b. John considered Peter intelligent (for years).
c. John thought Peter to be loyal (for years).
... Transitive Ground UOCs in Latin The following are examples of Ground
UOCs in Latin:
() Latin; Ov. Met. ,
Uberaque e-biberant
breast.ACC.PL=and out-drink.PLUPRF.PL
avidi [ . . . ] nati.
eager.NOM.M.PL born.NOM.M.PL
‘And her babes had drunk her breasts to exhaustion.’
() Latin; Plin. Nat. ,
Dracones esse tantos ut totum
snake.ACC.PL be.INF so_many.ACC.PL that whole.ACC.M.SG
sanguinem capiant, itaque elephantos ab iis
blood(M)ACC.SG take.SBJV.PL therefore elephant.ACC.PL by them.ABL
e-bibi.
out-drink.INF.PASS
‘That the snakes are so large that they can take all the blood, and therefore the
elephants are drunk dry by them’.
Note that these examples involve the prefixed verb e-bibo “out-drink”, which we have
already seen heading Figure UOCs (see section ...). Ground UOCs with e-bibo
“out-drink” present accusative objects referring to the container of the liquid, instead
Latin as a satellite-framed language
of the liquid itself. In the examples above the object is ubera ‘breasts’ and elephantos
‘elephants’, respectively, and, although they correspond to the Ground in the event
schema, they can be said to license a complete affection interpretation, as is clear
from the translation of ()—see the relevant entries for this verb in Lewis
and Short (), and Gaffiot (). It is worth observing that while simple bibo
‘drink’ may be used with container-naming objects, as in () below, I have
not found any such example (in a search of all the occurrences of simple bibo
‘drink’ in the Antiquitas corpus) with a non-standard container, such as those in
() and ():
() Latin; Plaut. Stich.
Vide quot cyathos bibimus.
see.IMP.SG how_many goblets.ACC drink.PRF.PL
‘See how many goblets we have drunk.’
This fact strongly suggests that cases such as () involve a metonymical reading of
the object, precisely because it refers to a canonical container holding a standard
quantity of liquid. The predicates in () and (), however, do not involve
metonymy. Neither the breasts nor the elephants are taken as standard measures
for the liquids they contain, nor are they, for that matter, conceived of as containers
of milk and blood, respectively. Rather, they seem to be really interpreted as the
Grounds in the motion schema. This is unexpected in, for instance, Lehmann’s
(: ) account of preverbation, in which the subject and the object of an
intransitive and a transitive unprefixed verb, respectively, emerge as the locatum—
in our terms, the Figure—of the preverb when the verb is prefixed.
The difference between Figure UOC e-bibo “out-drink” and Ground UOC e-bibo
is easily grasped: in () and (), for instance, the objects are not brought out or
made to disappear by virtue of a drinking event, as is the case in instances of e-bibo
“out-drink” in Figure UOCs. I repeat an example from section ... for the sake of
comparison:
() Latin; Petr. Sat. ,
Tantum medicamentum e-bibisti?’
so_much.ACC medicine.ACC out-drink.PRF.SG
‘Have you drunk up so much medicine?’
In this example, the prefix e- ‘out’, encoding, as was discussed in section ..., a
‘state of disappearance’, is predicated of tantum medicamentum ‘so much medicine’,
which is, thereby, a Figure. This is clearly not the interpretation of ubera ‘breasts’ and
elephantos ‘elephants’ in () and (), respectively.
The same difference is appreciated when contrasting the Figure UOC ab-luo “away-
wash” of (), repeated here as ()a, with the Ground UOC ab-luo of ()b:
S-framed constructions in Latin
Dracones Voice’
Voice vP
v PathP
Place ex
This analysis proposes the same basic argument structure for the second ob-duco
“against-lead” and the first obduco of (), where the Figure emerges as an accusative
DP and the Ground is the root of the prefix ob- ‘against’. The difference lies in the fact
that in () no DP has been merged as Spec-Place, so there is no Figure argument in
the argument structure—although an ablative adjunct such as spinis ‘thorns’ in ()
32
Zeller (a: ) also observes the fact that a same particle-verb in German may head a predicate
where the object is interpreted as Figure or a predicate where it is interpreted as Ground. This is the case
with German ab-spülen “off-rinse”, used either as a removal verb (‘rinse off the grease’) or as a change-of-
state verb (‘rinse off the dish’). See also McIntyre ().
S-framed constructions in Latin
could be interpreted as a ‘demoted figure’. This is why the Ground (tuum sepulcrum
‘your grave’) emerges as accusative and is moreover understood as a Measurer for the
event (since it is pulled up by Path onto its specifier), providing thereby what Van Laer
interprets as a ‘covering’ sense in the resulting predicate. In Van Laer’s () analysis
both senses of ob-duco “against-lead” remain unrelated.
There are cases of Ground UOCs where the Ground is not physical, but meta-
phorical. Thus, for instance, we find predicates of utterance where the addressee is
realized as the accusative object. The verb is marked with the prefix ad- ‘at’:
() Latin; Verg. Aen. ,
Talibus ad-fata Aenean.
such.ABL.PL at-say.PTCP.PFV.NOM.F.SG Aeneas.ACC
‘Having addressed Aeneas with those words’.
() Latin; Plaut. Cist.
Ad-hinnire equolam possum ego hanc
at-neigh.INF mare(F)DIM.ACC.SG can.SG I this.ACC.F.SG
‘I can well neigh at this little mare myself.’
() Latin; Plaut. Amph.
Obsecro ut [ . . . ] liceat te al-loqui
beseech.SG that be_allowed.SBJV.SG you.ACC at-speak.INF
‘I beseech you to let me address you.’
These cases help us further illustrate how the syntactic structure dictates the number
and interpretation of the arguments of a verb, overriding whatever information is
contained in the encyclopaedic entry of its root. In particular, if it is assumed that the
prefixed predicates in () to () involve movement of the Ground DP to Spec-
Path position, there is predictedly at most and at least one overt argument per
prefixed predicate, since, on the one hand, there is no position left for any other
argument in PlaceP (since, by hypothesis, Spec-Place is not filled), and, on the other
hand, each PathP must have its specifier. This is what happens in the above examples,
with only an accusative object naming the addressee, and the utterance argument
being expressed, at most, as an instrumental adjunct in the ablative, as is the case of
talibus ‘with such (words)’ in (). Descriptively, it could be said that the utterance
argument is ‘demoted’ to adjunct-status.33 The unprefixed counterparts to al-loquor
“at-speak”, af-for “at-say”, or ad-hinnio “at-neigh” display, as expected, a different
syntax. Notably, they cannot link an addressee as object. They are either unergative
(see ()a, () and ()a), or take an accusative object, which is, however,
33
These ad-verbs are strikingly similar to an-prefixed verbs in German, like an-lügen “at-lie” ‘lie to’ or
an-motzen “at-whinge” ‘whinge to’, discussed by Stiebels () and McIntyre (), where the addressee
is expressed as the accusative object DP.
Latin as a satellite-framed language
interpreted as the utterance. In either case, they may optionally appear with a dative
or a PP expressing the addressee (see ()a for the former option and ()c and
()c for the latter):
() Latin; simple for ‘say’
a. Liv. , ,
Mihi ita Iuppiter fatus est.
me.DAT thus Jupiter.NOM say.PRF.SG
‘Jupiter has talked to me thus.’
b. Verg. Aen. ,
Ea fatus erat.
those_things.ACC say.PLUPRF.SG
‘He had said that.’
c. Cic. Tim.
Ad eos is deus [ . . . ] fatur haec.
at them.ACC that.NOM god.NOM say.SG this.ACC.N.PL
‘To them that god says these words.’
() Latin; simple hinnio ‘neigh’, Ps. Apul. Herm.
Proprium est equi hinnire.
typical.NOM.N.SG is horse.GEN neigh.INF
‘It is typical of the horse to neigh.’
() Latin; simple loquor ‘speak’
a. Ov. Rem.
Illa loquebatur.
She.NOM speak.IPFV.SG
‘She was speaking.’
b. Cic. Tusc. , ,
Pugnantia te loqui non vides?
contradiction.ACC.PL you.ACC speak.INF not see.SG
‘Are you not aware that you are saying contradictions?’
c. Ov. Pont. , ,
Certus eras [ . . . ] numen
sure.NOM.M.SG be.IPFV.SG divine.ACC
ad Augustum [ . . . ] loqui.
at Augustus.ACC speak.INF
‘You were resolute to speak to divine Augustus.’
It seems, once again, that verbs (in fact, roots) have little to say on the realization of
arguments. Rather, it is the syntactic structure that determines the number and
quality of the arguments.
S-framed constructions in Latin
I finish this section by reconsidering the examples of CDMCs with a prefixed verb
and a directional accusative-marked DP, as seen in section ..:
() Latin; Tac. Ann. ,
Novissimos in-currere.
rear.ACC in-run.PRF.PL
‘They charged against the rear.’
As was shown in section .., the accusative DP in these predicates seems to be an
argument, since it can become a passive subject. These cases are amenable to an
analysis in terms of a Ground UOC in which the accusative DP is merged as Compl-
Place and in which there is no Figure argument:
() Analysis of ()
VoiceP
pro Voice’
Voice vP
v PathP
Place in
v PathP
v plu parietes Path’
Path PlaceP
Place parietes
Place per
of-location (COL) alternant, the object is the thing being located in a place, which is
expressed via a PP. In ()b, the change-of-state (COS) alternant, the syntax of
those two participants in the event is reversed, so that the object expresses the
location and the PP encodes the thing being moved. Moreover, it has very often
been observed that while ()b entails that the basket ends up full of apples, ()a
does not. ()b exhibits, therefore, the phenomenon known as ‘holistic effect’.34
Many studies have been devoted to the LA and a division can be made into two
basic types of approach. On the one hand, there are approaches where the COS
alternant is derived from the COL alternant, which is, thus, more ‘primitive’ (see
Larson ; Damonte ; Wunderlich , among others). These approaches,
based on classical theta-roles such as Theme and Location, aim at preserving a
privileged linking relation between the Theme role (apples in ()) and the syntactic
position of the object. On the other hand, there are approaches where the alternation
is not seen as a phenomenon to be explained in terms of a derivational relation
between both alternants (see Pinker ; Mulder ; Baker ; Mateu c;
Borer b, among others). These approaches adopt a significantly more abstract
view of theta-roles, which allows them an isomorphic mapping between the object
and its thematic interpretation without resort to a derivational mechanism. In
particular, for these approaches both apples and the basket receive the same ‘theta-
role’, so it comes as no surprise that they are both realized as objects. Here I will
follow a hybrid approach to the LA: although I believe that the non-derivational
approach is basically right for most cases of the LA, I will propose that some instances
of the LA do involve, at least in Latin, the derivation from one alternant to the other.
... The LA and the s-/v-framed distinction Importantly, the LA is the locus of
cross-linguistic variation, being quite rare in v-framed languages. Specifically, COL
alternants are hard to obtain in these languages—see Mateu (c) for Catalan and
Spanish, Rosselló () for Catalan, and Lewandowski () for a quantitative
study of Spanish. I illustrate this cross-linguistic asymmetry by the following
failed alternations in Catalan, which are perfectly acceptable in English. Note that
the a-sentences are COL alternants and the b-sentences are COS alternants:
() Catalan ruixar ‘spray’
a. *En Marc va ruixar aigua sobre la planta.
the Marc PRF.SG spray.INF water on the plant
‘Marc sprayed water onto the plant.’
b. En Marc va ruixar la planta {d’/amb} aigua.
the Marc PRF.SG spray.INF the plant of/with water
‘Marc sprayed the plant with water.’
34
See Anderson (); Dowty (); and Beavers (), among others. See also section ....
Latin as a satellite-framed language
35
As pointed out by Pinker (), Marantz (), and Borer (b), among others, the verb fill is a
conspicuous example of a verb that does not admit a COL alternant, even though English is an s-framed
language: *Fill water in(to) the glass. I assume that fill is an idiom in English, relating a root to a specific
position (Compl-Place). This is suggested by the fact that other languages do admit fill as a manner verb,
e.g. German füllen (Ambridge and Brandt ). On the other hand, acquisition studies such as that of
Gleitman and Landau () have shown that ‘English three year olds say “Fill water into the glass” almost
per cent of the time’ (Gleitman and Landau : ). This is compatible with the idea that roots are
freely inserted into structures, as shown by the more liberal grammar of children, and also that in the adult
grammar of English fill is fixed as a result verb through idiomaticity.
S-framed constructions in Latin
PRO Voice’
Voice vP
v PathP
stercus Place’
Place pratum
Place in
b. VoiceP
(tu) Voice’
Voice vP
v PathP
pabulum Path’
Path PlaceP
pabulum Place’
Place sparg
S-framed constructions in Latin
Note that in this non-derivational analysis the LA presented by spargo ‘scatter’ boils
down to the possibility of associating the same root with different positions of one
basic abstract configuration encoding an externally originated transition. Specifically,
in the COL alternant the root SPARG ‘scatter’ is merged as an adjunct to v, and is
interpreted, consequently, as a Co-event of the transition (change-of-location) event.
The COL alternant is, according to this analysis, an s-framed construction. In section
... I will argue that this type of unprefixed COL alternant involves, in fact, a
verbal root merged as Compl-Place. In the COS alternant the root is merged at
Compl-Place, and is interpreted as a Terminal Ground, as the final state of a
transition (change-of-location) event. As regards the object, it is a Figure in both
cases, since it is first merged at Spec-Place. However, since in the COL alternant it
appears in a predicative relation with a location, codified by in pratum ‘onto the
meadow’—with a root IN ‘in’ specifying the head Place and inducing a spatial reading
thereof—it is interpreted as an entity that changes location. By contrast, in the COS
alternant it holds a predicative relation with the verbal root, and is therefore
interpreted as an entity that enters into a specific state (a state of being ‘scattered’,
identified with SPARG). Observe, importantly, that I am positing the projection of a
PathP for both COL and COS alternants, and that in both cases the Path head raises
the nearest DP in its c-commanding domain, the Figure, to Spec-Path, where it is
interpreted as a Measurer. This means that in both cases the so-called holistic effect
must emerge, as seems to be the case: in ()a stercus ‘manure’ measures out the
event as much as pabulum ‘fodder’ does in ()b. This is in tune with Dowty’s
() observation that the objects of either COL or COS alternants are interpreted
as Incremental Themes, and that, if possessing the appropriate quantificational
properties, they might induce telicity in the predicate:36
() Dowty (), in Baker (: )
a. John sprayed this whole can of paint onto subway cars in an hour.
b. John sprayed this wall with paint in an hour.
Thus, the fact that pratum ‘meadow’ in ()a is not interpreted holistically (the field
need not end up covered with manure) is a syntactic effect: it cannot raise to Spec-
Path, and, hence, cannot be interpreted as a Measurer.
Note, finally, that I am treating the ablative amurca ‘dregs of oil’ in the COS
alternant of ()b as an adjunct to vP, as also proposed by Rappaport and Levin
(), Mateu (c), and Borer (b).37
36
See also Pinker (: ) and Borer (b: ). The latter capitalizes on this fact to show that in
both COL and COS the object is a Subject-of-quantity, sitting at Spec-AspQ (see section ..).
37
By contrast, and specifically for Latin, Pinkster (: ) considers these ablatives as arguments
(complements in his terminology).
Latin as a satellite-framed language
... The LA and prefixation. The heterogeneity of the LA The LAs shown in
examples () to () do not exhaust the exploration of the LA in Latin. Rather, it
has been observed (Hofmann and Szantyr ; Lemaire ) that this form of
argument structure alternation is very frequently mediated through prefixation. In
the following sections I capitalize, therefore, on the patterns of prefixation shown
by both alternants in the LA in Latin, and put them in relation both to other
constructions of the language and to similar patterns in other languages.
I purport to show that the different morphological manifestations of the LA in
this language suggest that it might be a rather heterogenous phenomenon, calling
for a non-uniform account.
One first prefixal pattern shown by the LA in Latin involves the presence of a prefix
in the COL alternant. The verbs laedo ‘hit, harm’ and quatio ‘shake, agitate’ illustrate
this pattern (I present the COL alternant first):38
() Latin in-lido “in-hit” ‘thrust against’ and laedo ‘hit’
a. Verg. Aen. ,
Notus [naves] in-liditque vadis.
south_wind.NOM ship.ACC.PL in-hit.SG=and sandbank.DAT.PL
‘The south wind thrusts the ships against the sandbanks.’
b. Plaut. Bacch.
Lembus ille mihi laedit latus.
boat.NOM that.NOM me.DAT hit.SG side.ACC
‘That boat hits my side.’
() Latin; quatio ‘shake, agitate’ and in-cutio “in-shake” ‘stamp against’
a. Quint. Inst. , ,
Terrae pedem in-cutere.
earth.DAT foot.ACC in-shake.INF
‘To thrust the foot against the earth’.
b. Hor. Carm. , ,
Terram quatiunt pede.
earth.ACC shake.PL foot.ABL
‘They shake the earth with their feet.’
38
Laedo ‘hit, harm’ and quatio ‘shake, agitate’ can be said to enter, in () and (), what Levin and
Rappaport Hovav () call the with/against alternation, exemplified below, which involves impact verbs:
(i) Levin and Rappaport Hovav (: )
a. Kerry hit the stick against the fence.
b. Kerry hit the fence with the stick.
S-framed constructions in Latin
The prefixed counterparts in-cutio “in-shake” ‘stamp against’ and in-lido “in-hit”
‘thrust against’ are not found as COS alternants (Lewis and Short ). In the
analysis given here, the prefix corresponds to a root merged as an adjunct to Place,
where it is thus interpreted. The verbal root is merged as an adjunct to v, specifying
the kind of transition undergone by the Figure.
As shown by Lemaire (), many cases of the LA involve the same prefix for
both alternants. I illustrate with circum-icio “around-throw” ‘surround’ and in-duco
“in-lead” ‘smear’:
() Latin circum-icio “around-throw” ‘surround’
a. Liv. , ,
Fossam [ . . . ] uerticibus iis, quos
ditch.ACC peak(M)DAT.PL those.DAT.M which.M.ACC.PL
in-sederant, circum-iecere.
in-sit.PLUPRF.PL around-throw.PRF.PL
‘They put a ditch around the peaks where they had settled down.’
b. Tac. Ann. , ,
Planitiem saltibus circum-iectam.
plain(F)ACC forest.ABL.PL around-throw.PTCP.PFV.ACC.F
‘A plain surrounded by forests’.
() Latin in-duco “in-lead” ‘smear’
a. Cels. ,
Ulceri medicamentum [ . . . ] in-ducatur.
ulcer(N)DAT.SG medicament.NOM.N.SG in-lead.SBJV.PASS.SG
‘Let the medicament be smeared into the ulcer.’
b. Plaut. Most.
Postes [ . . . ] sunt in-ducti pice.
doorpost.NOM.PL be.PRS.PL in-lead.PTCP.PFV.NOM.M.PL pitch.ABL.SG
‘The doorposts have been smeared with pitch.’
In the COS alternants of these instances of the LA, the objects, which happen to be
passive in both examples, hold a Ground semantic relation with the prefixes. Thus, in
()b the forests (saltibus) are around (circum-) the plain (planitiem), and in ()b
the pitch (pice) is smeared into (in-) the doorposts (postes). Thus, these cases of COS
alternants can be treated as Ground UOCs, with no DP merged at Spec-Place and
with the Ground raising to Spec-Path:
Latin as a satellite-framed language
v PathP
v ic
planitiem Path’
Path PlaceP
Place planitiem
Place circum
Therefore, in these cases of COS alternants endowed with a spatial prefix, I argue for
a derivational approach to the LA: these COS alternants are derived from structures
where the object is first merged as a Ground and there is no Figure merged at Spec-
Place.
Finally, I point out that many verbs that are prefixed with co(m)- ‘together’ are
only interpreted as COS alternants. Thus, in the following examples the object
(passivized or not)—ora ‘face’ and me ‘me’, respectively—seems to be interpreted
as an entity that changes state through a locating event (of covering with make-up or
tears, respectively):
() Latin; Ov. Rem.
Con-linit ora venenis.
together-smear face.ACC make-up.ABL.PL
‘She covers her face completely with make-up.’
() Latin; Cic. Planc.
[Me] con-spersitque lacrimis.
me.ACC together-scatter=and tear.ABL.PL
‘And he covered me with tears.’
By contrast, the absence of the com- ‘together’ prefix licenses a COL reading. Thus,
the following predicates feature an unprefixed verb and present a COL reading and
a COL syntax. Thus, the object (again, passivized or not) is interpreted as a
Figure and per corpora ‘on the whole body’ and in pratum ‘onto the field’ are the
Grounds:
() Latin; Ov. Medic.
Medicamina [ . . . ] lini per corpora possint.
makeup.NOM smear.INF.PASS through body.ACC can.SBJV.PL
‘Such a makeup as may be smeared on the body’.
S-framed constructions in Latin
39
Lemaire (: ) also observes a contrast between pairs like con-scribo “with-write” ‘cover with
inscriptions’ and in-scribo “in-write” ‘inscribe, write in(to) or upon’, interpreting an opposition between a
‘contact’ sense of com- and an ‘insertion’ sense of in-, which introduces the sense of insertion. On the other
hand, Meillet (: ff.) and Barbelenet (: –, –) had already noticed a grammatical-
ization of the prefix com- ‘together’ when they argued that it constituted a morphological way to induce a
‘perfective’ interpretation in an otherwise ‘durative’ verb (as in specio ‘watch’ / con-spicio “with-watch”
‘spot’). See also Moussy ().
Latin as a satellite-framed language
pro Voice’
Voice vP
v PathP
v lin ora Path’
Path Place
ora Place’
Place com
Note that, as usual, the merging of PathP as a sister to v brings about movement of
the highest DP, ora ‘face’, onto its specifier. A paraphrase for this predicate would be
‘to affect the face completely through a making-up event’.
That these com-prefixed COS alternants are s-framed constructions is not surpris-
ing when we consider that they mirror analogous predicates in other languages
claimed to be s-framed. Thus, in the following sentences the particles be (Dutch),
be (German), and meg (Hungarian) induce a complete affection interpretation:
() Dutch; Hoekstra and Mulder (: )
Hij be-hing de muur met posters.
he be-hang.PST.SG the wall with posters
‘He covered the whole wall with posters.’
() German; Wunderlich (: )
Er be-giesst die Blumen mit Wasser.
he be-pour.SG the.ACC.PL flower.ACC.PL with water.DAT
‘He waters the plants (with water).’
() Hungarian; Ackerman (: )
A paraszt meg-rakta a szekeret (szénával).
the peasant meg-load.PST.SG the cart.ACC hay.INSTR
‘The peasant loaded the cart full with hay.’
Specifically for Dutch, Hoekstra and Mulder (: –) and Mulder (:
–) claim that the prefix be-, inducing complete affection, is in fact a predicate
heading a Small-Clause-like structure, since, it happens to be in complementary
distribution with a resultative AP (vol ‘full’ in the example):
() Dutch; Mulder (: )
*Hij be-hangt de muur vol me foto’s.
he be-hangs the wall full with photos
S-framed constructions in Latin
As will become clear in Chapter , I cannot apply this test to Latin, since Latin does
not license complex AP resultative constructions. However, com- ‘together’ can
change the argument structure properties usually displayed by the unprefixed verb,
and, in that sense, it is amenable to an analysis along the lines of those proposed
above for other prefixes which induce changes in argument structure. I underpin this
claim with the contrast between mingo ‘piss’, an intransitive creation verb (see ()
a) and com-mingo “together-piss” ‘piss all over’ (see ()b):
() Latin mingo ‘piss’ and com-mingo “together-piss” ‘piss all over’
a. Mart. , ,
Minxisti currente semel, Pauline, carina.
piss.PRF.SG run.PTCP.PRS.ABL.F once Paulinus.VOC boat.ABL
‘You pissed once, Paulinus, while the ship was sailing along.’
b. Hor. Sat. , ,
Com-minxit lectum.
together-piss.PRF.SG bed.ACC
‘He pissed the bed.’
As usual, I treat the unselected object lectum in ()b as a Figure, while the prefix
originates as a predicative root in Compl-Place and the verbal root is an adjunct to v:
() Analysis of ()b
[VoiceP pro [Voice’ Voice [vP [v v MING] [PathP lectum [Path’ Path [PlaceP lectum
[Place’ Place COM]]]]]]]
From this discussion a scenario emerges in which the LA might be more heterogenous
than previously considered. Specifically, COS alternants may respond to different
syntactic strategies based on the type of element merged as the Terminal Ground at
Compl-Place. They can be change-of-state predicates with the verbal root merged as a
Terminal Ground (see ()b), they can correspond to Ground UOCs, with the object
first merged as a Terminal Ground (see ()b), and they can correspond to predicates
with the prefix com- ‘together’, inducing a complete affectedness semantics, merged as
a Terminal Ground (see ()). In the second case, crucially, the COS alternant can be
said to derive from a basically COL structure that lacks, however, a Figure. I summarize
the scenario for the LA in both v- and s-framed languages in the table below:40
40
Hofmann and Szantyr (: ) document a kind of the LA built around adjectival predicates:
(i) Latin; based on Hofmann and Szantyr (: )
a. flores plenae in campo (COL alternant)
flower(F)NOM.PL full.NOM.F.PL in field.ABL
b. campus floribus plenus (COS alternant)
field(M)NOM flower.ABL.PL full.NOM.M
Latin as a satellite-framed language
.. Pseudoreversatives
The last constructions I would like to deal with are the ones McIntyre (: )
calls Pseudoreversatives, which, to my knowledge, have not been dealt with before in
the literature on Latin.42 These are constructions where ‘the result expressed or
implied by the base verb gets reversed by adding a particle which contradicts this
result’ (McIntyre : ).43 The following German particle-verbs illustrate:
() German; McIntyre (: )
a. aus-parken
out-park.INF
‘Drive (a car) out of a parking space’.
b. ab-schwellen
down-swell.INF
‘Swell down, become less swollen’.
c. los-binden
free-tie.INF
‘Untie (a horse, etc.)’.
These constructions once again exemplify the s-framed pattern: the verb indicates the
nature of the process involved and a morphologically different element encodes the
These examples show that plenus ‘full’ could be predicated both of the entity that is full of something (see
(i)b) and of the matter or objects of which something is full (see (i)a). I leave this striking kind of the LA for
future research.
41
See Munaro () for Italian cases of the LA involving a contrast between an unprefixed verb and a
prefixed verb.
42
Although, for a diachronic remark on cases like aperio ‘open’/operio ‘close’, which had arguably been
cases of Pseudoreversatives, see Turcan (: –).
43
See also Stiebels ().
S-framed constructions in Latin
Core Schema. Thus, in ()a the conceptual scene evoked is the same as that evoked
by the verb parken ‘park’, the driving of a vehicle, but the result part of the event
usually entailed by parken ‘park’ is missing: the car is not in the parking space by the
end of the event. The addition of the particle aus- ‘out’ imposes a different result
state: the car ends up out (of the parking space). Pseudoreversatives are, therefore, a
particularly interesting probe into the nature of the semantic contribution of the verb
in s-framed constructions: it is truly understood as an adjunct, a modifier of the
event, the result being codified by an independent element (the particle, in the
examples above). Unsurprisingly, Latin features Pseudoreversatives, as exemplified
below:
() Latin; Verg. Aen. ,
Quid moror? an mea Pygmalion
what.ACC delay.SG whether my.ACC.N.PL Pymalion.NOM
dum moenia frater de-struat [ . . . ]?
until wall(N)ACC.PL brother.NOM down-build.SBJV.SG
‘What am I waiting for? Maybe for my brother Pygmalion to destroy my
walls?’
() Latin; Plaut. Curc.
Valetudo de-crescit, ad-crescit labor.
health.NOM down-grow.SG at-grow.SG work.NOM
‘Health wanes; work increases.’
() Latin; Ov. Met.
Dis-iunxisse iuvencos.
asunder-yoke.INF.PFV oxen.ACC
‘Having unyoked the oxen’.
() Latin; Ov. Fast. ,
Dis-suto [ . . . ] sinu.
asunder-sew.PTCP.PFV.ABL.M pleat(M)ABL
‘With an unsewn pleat’.
() Latin; Plaut. Cist.
Ex-pungatur nomen, nequid debeam.
out-puncture.SBJV.PASS.SG name.NOM nothing.ACC owe.SBJV.SG
‘Let my name be erased (from the register of debtors), so that I’m left with no
debts.’
() Latin; Colum. ,
Ne ventis [pampini] ex-plantentur.
lest wind.ABL.PL shoot.NOM.PL out-plant.SBJV.PASS.PL
‘Lest the vine shoots be uprooted by the wind’.
Latin as a satellite-framed language
In all these examples the result inferred from the unprefixed verb is superseded by
that conveyed by the prefix. Thus, in () nomen ex-pungo “name out-puncture”
refers to the action opposite to nomen pungo ‘puncture a name’, that is, ‘write a name
by puncturing’, on a wax tablet with a sharp instrument. The name is, in effect, ‘taken
out of the tablet’, and this is conveyed by ex- ‘out’. The effect expressed by ex-pungo
“out-puncture” is, thus, that of erasing.
I propose that these constructions receive the same analysis as Figure UOCs. They
involve a PathP with a prefix encoding the result, and a root adjoined to v. The object
is merged as Spec-Place, and is interpreted as a Figure. In the following example
involving de-struo “down-build” ‘destroy’ (cf. the German literal correspondence
ab-bauen), the walls (mea moenia) are predicated to end up down (de-). The
Figure raises then to Spec-Path and is interpreted as a Measurer of the event:
() Analysis of ()
VoiceP
v PathP
Path PlaceP
44
As regards de- ‘down, downwards’, Brachet (: ff.) points out that it is found as a pure
‘opérateur d’inversion’. However, he acknowledges (Brachet : ff.) that in the first attestations of
S-framed constructions in Latin
the reversative interpretation is a secondary effect derived from a clash between the
semantics of the prefix and the semantics of the verb, as has been shown for ().
The examples () t () are illustrative of the fact that the prefixes do have a
locational meaning. In () the verb dis-suo “asunder-sew” is secondarily interpreted
as the opposite of suo ‘sew’, but the final state encoded by the prefix is specifically that
of separation (of two pieces of fabric, in this case). This semantic nuance is different
from that conveyed by the prefix ex- ‘out’ in ex-planto “out-plant” (see ()),
where the final result is for the plant to be out of the earth. Note, finally, the contrast
obtained by the combination of two different prefixes with the same verb in ().
Pseudoreversatives make it particularly evident that some of the meaning compo-
nents traditionally attributed to roots, such as ‘state’ are, in fact, derived from the
structure. Thus, run-of-the-mill change-of-state verbs like iungo ‘yoke’ or planto
‘plant’ simply cannot be interpreted as such if their root is not inserted as Compl-
Place. In theories such as Harley’s () and Levinson’s () roots are typed
depending on what they encode: entity, state, or event. For instance, Levinson (:
), following Harley’s () classification of roots, takes a root such as OPEN as
being typed as <ss, t>, that is, a state. This explains the adequacy of this root in
change-of-state predicates:
() The archaeologist opened the sarcophagus.
However, a typing approach such as this one, when applied to cases of verbal
elasticity such as the one at hand, is forced to propose different groups of homoph-
onous roots distinguished by the semantic type. For instance, the root IUNG ‘yoke’ in
() must be of type <e, t>, entity (akin to that of the cognate noun iugum ‘yoke’),
since the (end) state in that predicate is codified by the prefix dis- ‘asunder’, and not
by the root. However, in the following example the root is interpreted as a final state
(that of ‘being yoked’), which would require type <ss, t>:
() Latin; Hyg. Fab. , ,
Equum cum boue iunxit ad aratrum.
horse.ACC with ox.ABL yoke.PST.SG at plough
‘He yoked the horse with the ox to the plough.’
The scenario in which roots, like DP arguments, receive an interpretation dictated by
their position in the configuration (see section ...) does away with this redun-
dancy problem.
de-prefixed verbs exhibiting a reversative meaning, the prefix retains the ‘downward’ nuance: de-scendo
“down-ascend” ‘go down’ (from scando ‘climb’), de-cresco “down-grow” ‘diminish’ (from cresco ‘grow’),
de-molior “down-construct” ‘demolish’ (from molior ‘construct’).
Latin as a satellite-framed language
. Summary
In this chapter I have shown that Latin is an s-framed language, in Talmy’s ()
sense, since the Core Schema, that is, the component specifying a transition into a
final state or location, and the eventive component of transition events are realized
independently. In addition, I have endeavoured to show the adequacy of a neo-
constructionist model in dealing with the constructions which make Latin an s-
framed language. I have introduced Talmy’s theory, and I have adapted it to the
theory introduced in section ., introducing a correspondence between the semantic
components in Talmy’s theory of transition events and the syntactic-semantic terms
of my theory. I have shown that the s-/v-framed distinction can be explained as a
result of a different interpretation of the structure at PF: in v-framed languages, v and
Path must be strictly adjacent to each other, which disallows the linear intervention
of any root previously adjoined to v. This mechanism explains why v-framed
languages do not feature constructions involving a manner-naming verb and an
expression encoding the Core Schema. In s-framed languages there is no such
adjacency requirement between v and Path, and, hence, v can be associated with a
root merged as an adjunct. This analysis, where v-framed languages are more
complex than s-framed ones with respect to the PF derivation, makes the welcome
prediction that s-framed languages allow v-framed constructions, that is, predicates
where the verb encodes the Core Schema: there is nothing in s-framed languages
precluding these constructions. I have provided an overview of the expression of
directionality in Latin, in the form of a prefix, a PP, or a DP, or a combination of the
former and the latter. I have pointed out that APs cannot express the PathP in Latin.
I have presented the evidence that Latin is an s-framed language by approaching a set
of constructions that conform to the s-framed schema: CDMCs, Figure UOCs,
Ground UOCs, constructions involved in the LA, and Pseudoreversatives. All these
constructions have been argued to involve a verbal root merged as an adjunct to v
and interpreted, consequently, as a Co-event. In turn, the PathP is expressed through
an independent element. In the discussion of all these constructions I have tried to
show how the facts naturally derive from a neo-constructionist account, where it is
the syntactic structure, independently of the roots inserted therein, that determines
the structural semantics and the argument structure properties of the constructions.
5
resultatives, the former being found only in s-framed languages, the latter being
found in both s- and v-framed languages; and, finally, by discussing the situation
aspect that complex resultative constructions usually involve.
1
The term resultative construction has almost always been applied to complex resultative constructions
where the XP expressing the result state is an AP—see Halliday (); Simpson (); Levin and
Rapoport (); Hoekstra (); Carrier and Randall (); Levin and Rappaport Hovav (,
); Neeleman and van der Koot (); Mateu (); Boas (); Kratzer (); and Tomioka
(), among others. Crucially, I use the term in a wider sense.
The non-existence of complex adjectival resultatives in Latin and Slavic
() Catalan
La Sue deixà la taula neta.
the Sue leave.PRF.SG the table clean
‘Sue made the table clean.’
() Catalan
La Sue netejà la taula.
the Sue clean.PRF.SG the table
‘Sue cleaned the table.’
In the discussion central to this chapter I will focus almost only on complex
resultative constructions, but see section .. and, particularly, section .., for
simple resultative constructions in Latin based on a light change-of-state verb and an
AP, as in ().
... Strong and weak resultative constructions A second initial clarification that
needs be made is about the difference between so-called strong and weak (com-
plex) resultative constructions. Importantly, Washio (), in his comparison of
English and Japanese adjectival resultative constructions, makes a distinction
between these two types of resultative constructions, illustrated by the following
examples:
of ()b entails the attainment of a result state, namely, that of being painted, and the
AP blue is a specification of that result state. Washio observes that the adjectival
resultatives allowed in Japanese are always of the weak type (see the Japanese rendition
of ()b in ()b), the strong type being disallowed (see the Japanese rendition of ()a
in ()a):
John-ga Voice’
Voice vP
v PathP
kabe-o Path
Path PlaceP
AP PlaceP
buruu-ni kabe-o Place’
Place nut
2
See Den Dikken (: ) for discussion of similar data from Jackendoff (: ).
3
Another problem for Borer’s account is the fact that the adjective and the verb may appear separate,
which is unexpected if they form a complex listeme. Moreover, it is not clear, within her account, why strong
resultative constructions are systematically ungrammatical in v-framed languages like Romance (Mateu ;
Acedo-Matellán and Mateu ). Indeed, if the peculiarity of these constructions boils down to the
embedding of two listemes, rather than one, within the functional structure, why are such languages unable
to combine them? Of course, the combination of two listemes into a complex one could be stipulated as
unavailable in their grammars. However, Borer rejects any account of cross-linguistic variation that is not
based on morphophonological properties of the functional lexicon (see Borer b: ff.).
Weak satellite-framed languages
the final resulting state encoded by the result predicate. This is illustrated by the
following example:
As Kratzer herself notes, the sentence above implies that ‘the boat will remain inflated
for a few hours’, and not that there will be any pumping-up event that will last two
hours. Accordingly, the adverbial für ein paar Stunden ‘for a few hours’ can be
claimed to be measuring the result state incarnated as PlaceP, where the root AUF
‘up’ is embedded as Compl-Place.
I want to make a final remark about the situation aspect of CDMCs, since I have
assumed that these constructions are also complex resultative constructions. As
I pointed out in section ..., CDMCs are usually taken to be telic. Cases in
point are the examples in ():
() Dutch and German; Randall, van Hout, Weissenborn and Baayen (: )
a. John is in twee seconden de kamer in gedanst.
John is in two seconds the room in dance.PTCP.PFV
‘John has danced into the room in two seconds.’
b. John ist in zwei Sekunden in-s Zimmer getanzt.
John is in two seconds in-the.ACC room dance.PTCP.PFV
‘John has danced into the room in two seconds.’
That the sentence in ()b does not involve an unergative predicate expressing non-
directed motion is further suggested by the fact that unquestionably non-directed
motion predicates, featuring either no spatial PP or a locative PP at most, present a
HAVE-auxiliary:
As I already pointed out in section .., the adjectival type is included by Talmy
() himself in the range of constructions possible in s-framed languages and
impossible in v-framed ones. This is illustrated in () through a contrast between s-
framed German and v-framed Spanish: while German encodes the resulting state of
the complex event as an AP, leaving the verb to express manner, Spanish encodes the
resulting state as the verb, and the manner has to be expressed as an adjunct. A literal
Spanish translation of () is not well-formed:
As was pointed out in section .., anyone acquainted with Latin recognizes the type
represented by () or () as not possible or general in this language, although it is
an s-framed one. That is, Latin does not seem to feature complex resultative con-
structions where the result predicate is an AP, as illustrated by the next made-up
example:
However, it is pretty evident that this type of resultative is not complex in the
sense intended in the present discussion, since it does not involve a co-event
leading to the causation of the final state described by the AP. Contrary to
Pinkster’s claim above, there seem to be other verbs in Latin licensing simple
resultative constructions: the prefixed verbs red-do “back-give” ‘return’ and re-
linquo “back-leave” ‘leave’, and the simple verb facio ‘do’. I will deal with simple
adjectival resultative constructions headed by these verbs and also by ago ‘lead,
drive’ in section ...
Weak satellite-framed languages
As pointed out by Embick (: ), the above restriction has to do with the fact
that the participle is in itself already resultative, that is, it presupposes a previous
event whose result it identifies.
Second, many of the examples provided by these authors are headed by a prefixed
verb. We have seen evidence in the preceding chapter that prefixed verbs in transition
predicates in Latin already encode the result of the event, specifically that the prefix
The non-existence of complex adjectival resultatives in Latin and Slavic
can be claimed to express it. Therefore, it seems that, if anything, the APs in the above
examples are further specifying a result state already encoded by the prefix, which
makes these examples cases of a different kind of adjectival resultative construction.
I will turn to the derivation of this kind of AP prefixed resultative at the end of section
... In light of these two empirical differences and the fact that, as pointed out by
the same authors, this phenomenon is very rare in prose, I consider that the above
evidence does not support a claim that complex AP resultatives are productive in
(Early and Classical) Latin.
To see whether () represents a general situation in Latin, that is, whether this
language in fact cannot generate complex AP resultative constructions, I have
performed a corpus search, based on the one designed by Boas () for English.4
Boas selected a set of adjectives recurrently used in the literature on resultatives as
result predicates, such as dead, clean, awake, etc. He then conducted a search to
find out which verbs were most often used in resultative constructions with those
adjectives.5
In applying this methodology to Latin, I have first established the correspondences
of the English adjectives in Boas’s set and then the correspondences of (some of) the
verbs he established as more collocative for each adjective, wherever possible.
I present below the list of the combinations I have searched for:
() Adjectives and verbs used in the search for adjectival resultatives in Latin
Adjectives Verbs
aeger ‘ill’ bibo ‘drink’
cassus/inanis/vacuus
bibo ‘drink’, haurio ‘scoop’, poto ‘drink’
‘empty’
calcitro ‘kick’, clamo ‘scream’, figo ‘prick’, grunnio
experrectus ‘awake’ ‘grunt’, osculor ‘kiss’, plaudo ‘clap’, quatio ‘shake’,
‘jerk’, terreo ‘induce terror’
amens/demens/insanus
clamo ‘scream’, loquor ‘talk’, strideo ‘yell’
‘insane, mad’
caedo ‘cut, knock’, calcitro ‘kick’, cudo ‘knock’, occido
mortuus ‘dead’
‘kill’, tundo ‘strike, knock’, verbero ‘smite’
4
Analogously, Whelpton () uses Boas’s () appendix of examples of adjectives, taken from the
British National Corpus, as the starting point for his own investigation of Icelandic resultatives (see
section ..).
5
See Boas (: ff.).
Weak satellite-framed languages
Adjectives Verbs
I have dismissed some of the verbs in Boas’s subcorpus. For instance, verbs such as
get, render, or make, which head simple resultative constructions. In some cases
I have added verbs which I imagined could be possible with the adjective. This is the
case of the verbs combining with aeger ‘ill’, or pinguis/opimus ‘fat’. The subcorpus
obtained was composed of all the sentences in which each adjective combined with at
least one of the verbs of the same row in the box. Despite the ample range of
adjectives and verbs used and their high absolute frequency in the Antiquitas corpus
(and in Latin in general), the results have been utterly negative. Therefore, my
conclusion is that Latin disallows this type of complex resultative construction.
The non-existence of complex adjectival resultatives in Latin and Slavic
This situation constitutes a puzzle within the perspective adopted here, where
adjectives are expected to be able to fulfil the role of result predicates in s-framed
languages in general. However, as I shall show in section .., Latin is not the only
s-framed language to ban the formation of AP resultatives.
.. No complex AP resultatives in Slavic
Slavic languages are considered by Talmy (: ) to be s-framed, since they
typically convey the Core Schema as an element different from the verb:
() Russian; Talmy (: )
Ptica [v]Core Schema -letela.
bird.NOM in -flew
‘The bird flew in.’
Gehrke (: ff.) disputes this claim by arguing that Slavic languages (Czech and
Russian, in particular), like v-framed languages, do not allow the integration of a non-verbal
predicate into an activity VP to derive an accomplishment structure. In these languages, the
author argues, resultativity is to be expressed in the verb, as shown with prefixed verbs.
However, by using a strictly Talmian perspective, this author misses the point that verbal
prefixes are precisely the kind of non-verbal predicates ‘integrated into an activity structure’
that are allowed in these languages (and, as I shall show, in Latin). At the basis of Gehrke’s
argument lurks the use of the word as a syntactic-semantic unity: Slavic (and Latin) can be
taken to be v-framed since the verb, that is, a word which may include, for instance, a verbal
prefix, is the privileged unit where resultativity is expressed. We note, however, that Talmy’s
() typology is constructed on considerations about morphemes, and not about words—
see section . for relevant discussion of the status of words as syntactic units. Gehrke’s
observation, however, does account for the fact that, as in Latin, when the result predicate is
an AP, the construction turns out to be ungrammatical:
() Russian; Strigin (: )
*Ona mylila men’a skolzkim.
she soap.PST me slippery
‘She soaped me slippery.’
Importantly, the contrast between () and () is not to be stated in terms of change
of location versus change of state. As it turns out, Russian (and Slavic, in general)
seems to succeed in mimicking typical adjectival resultative constructions found in
English, expressing a change of state, and even featuring unselected objects, with the
use of adpositional prefixes:
() Russian; Spencer and Zaretskaya (: )
a. Oni na-ezdili ètu dorogu.
they on-drive this road.ACC
‘They’ve made this road nice and smooth (by driving over it).’
Weak satellite-framed languages
6
The division, first proposed for Romance prefixes by Di Sciullo (, ), corresponds, roughly, to
that made between outer and inner prefixes (Padrosa and Markova ) and superlexical and lexical
prefixes (cf., for instance, Svenonius b and the other articles in the same volume on Slavic prefixes).
Weak satellite-framed languages
b. *Od(INT)-po(EXT)-stoupit.
from-a_little-step.INF
Syntactically, internally prefixed verbs show different argument structure properties
from those of their unprefixed counterparts. This is not the case with externally
prefixed verbs. For instance, in the first of the examples following, the internal prefix
makes the object obligatory. In the second example, the externally prefixed verb
allows the omission of the object, as does the unprefixed verb:
() Russian; Gehrke (: )
a. Na-pisat’P *(pis’mo).
on-write.INF letter.ACC
‘To write (up) *(a letter)’.
b. Po-pisat’P (pis’mo).
po-write.INF letter.ACC
‘To write (a letter)’.
Semantically, both internal and external prefixes induce (outer-aspectual) perfectiv-
ity, but only the former necessarily induce (inner-aspectual) telicity, according to
Gehrke (: ):
() Russian; Gehrke (: )
a. Ja na-pisalP pis’mo *(za) dve minuty.
I on-wrote letter.ACC in two minutes
‘I wrote a letter in/*for two minutes.’
b. On po-spalP (*za) dve minuty.
he po-slept in two minutes
‘He slept for/*in two hours.’
Finally, internal prefixes may trigger a special meaning of the base verb (see ()a),
while external prefixes only introduce aspectual (quantificational) modifications of
the whole event (see ()b):
() Serbo-Croatian; Arsenijević (: )
a. biti u-biti raz-biti pro-biti od-biti do-biti
beat in-beat around-beat through-beat away-beat to-beat
‘beat’, ‘kill’, ‘break’, ‘make a hole in’, ‘bounce’, ‘get’
b. kuvati na-kuvati iz-kuvati pro-kuvati pre-kuvati
cook on-cook out-cook through-cook over-cook
‘cook’, ‘cook many’, ‘cook all/fully’, ‘cook a bit’, ‘overcook’
There is reason to believe that the distinction between internal and external prefixes
holds in Latin too. Specifically, some prefixes act more like adverbs, rather than as
resultativity markers. For instance, the prefix sub- ‘under’ may be added to a simple
Latin and Slavic complex resultatives with an internal prefix
I will come back to this typology in section .. But first it should be shown that the
observation in () is empirically correct for Latin and Slavic. This is done in section ..
the aid of a prefix. The first two searches aim at finding constructions formed by a PP
and a surface-contact verb, in the first search, and a sound-emission verb, in the
second search, taking into account only unprefixed verbs. The type of resultative
constructions I am looking for in these searches is respectively illustrated by the
English constructions in () and ():7
() Rappaport Hovav and Levin (: )
Terry swept the crumbs into the corner.
() Folli and Harley (: )
Mary whistled Rover to her side.
The search involving sound-emission verbs, did not produce any results whatsoever,
confirming () for Latin. That involving surface-contact verbs yielded a few appar-
ent examples, those in () to ():
() Latin calco ‘tread, press’
a. Cato, Agr.
[Oleas] in orculam calcato.
olive.ACC.PL in vessel.ACC press.IPV.FUT.SG
‘Press [the olives] down into an earthenware vessel.’
b. Stat. Theb. ,
Clipeumque in pectora calcat.
shield.ACC=and in chest.ACC press.SG
‘He stands/presses his shield against his chest.’
() Latin tero ‘rub, grind; thresh’
a. Petron. ,
Sparserunt [ . . . ] ex lapide speculari pulverem
sprinkle.PRF.PL, out mica.ABL powder(M)ACC.SG
tritum.
grind.PTCP.PFV.ACC.M.SG
‘They sprinkled powder ground out of mica.’
7
The material used for these two searches is displayed in (i) and (ii), respectively:
(i) Latin; Search for complex PP resultatives with unprefixed surface-contact verbs
a. Verbs: calco ‘tread, press’, frico ‘rub’, rado ‘scrape, scratch; razor’, tergeo ‘wipe’, tero ‘rub, grind;
thresh’, verro ‘sweep’.
b. Prepositions: ab ‘off, away’, ad ‘at, beside, by’, de ‘downward; from, away’, ex ‘out of ’, in ‘in’.
(ii) Latin; Search for complex PP resultatives with unprefixed sound-emission verbs
a. Verbs: fremo ‘roar’, strideo ‘yell’, rideo ‘laugh’, sibilo ‘whistle’, latro ‘bark’, ululo ‘howl’, mugio
‘moo’, hinnio ‘neigh’, strepo ‘make a lot of noise’, grunnio ‘grunt’, rudo ‘bray’, balo ‘bleat’.
b. Prepositions: ab ‘off, away’, ad ‘at, beside, by’, de ‘downward; from, away’, ex ‘out of ’, in ‘in’.
Latin and Slavic complex resultatives with an internal prefix
8
The search for telic complex directed motion constructions with unprefixed and prefixed verbs
involved the following criteria:
(i) Prefixed and unprefixed verbs (‘p-’ represents any prefix):
(p-)ambulo ‘walk’; (p-)curro ‘run’; (p-)equito ‘ride’; (p-)fluo ‘flow’; (p-)gredior ‘walk, step’; (p-)labor
‘slip’; (p-)navigo ‘sail’; (p-)repo ‘crawl’; (p-)salio ‘jump’; (p-)volo ‘fly’
(ii) Telicity-signalling expressions (cf. Pinkster : ff., : ff.)
a. Adverbs
extemplo, repente, repentino, statim, subito or subitum, ‘suddenly’
b. Prepositions
intra ‘in’ (as in intra tres dies ‘in three days’)
c. Complementizers
simul ac, simul atque, ubi or ut primum, ‘as soon as’
d. Ablative forms of nouns and adjectives encoding periods of time
dies ‘day’, hora ‘hour’, nox ‘night’, mensis ‘month’, annus ‘year’, diurnus ‘of the day’, diutinus
‘lasting’, diuturnus ‘lasting’, nocturnus ‘of the night’, menstruus ‘which lasts a month’, menstrualis
‘which lasts a month’, annuus ‘which lasts a year’, annalis ‘of a year’, annualis ‘a year old’, aestas
‘summer’, hiems ‘winter’, ver ‘spring’, autumnus ‘autumn’, mane ‘morning’, vesper ‘evening’,
vesperus ‘of the evening’, calendae/kalendae ‘calends’, idus ‘ides’, nonas ‘nones’, lustrum ‘lustrum’,
meridies ‘noon’, vigilia ‘time of keeping watch by night’, hibernus ‘of the winter’, saeculum/
seculum/saeclum ‘century’, saecularis ‘of a century’, aestivus ‘of the summer’, aestivalis ‘of the
summer’, vernus ‘of the spring’, vernalis ‘of the spring’, autumnus ‘of the autumn’, autumnalis ‘of
the autumn’, horalis ‘which lasts an hour’, matutinus ‘of the morning’, postmeridianus ‘of the
afternoon’, vespertinus ‘of the evening’, spatium ‘time span’
e. Ablative form of adjectival suffixes indicating a period of time
-duus ‘of X days’, -ennius ‘of a year’, -noctius ‘of X nights’, -menstruus ‘of X months’, -menstris ‘of
X months’
Latin and Slavic complex resultatives with an internal prefix
b. Petron. ,
Statimque, ad re-ficiendum ignem in viciniam cucurrit.
at_once=and, at re-make.GER.ACC fire.ACC in neighbour’s.ACC run.PRF.SG
‘And immediately, he ran to the neighbor’s to kindle the fire.’
c. Suet. Otho ,
Ac repente omnes in palatium cucurrerunt.
and suddenly all.NOM.PL in palace.ACC run.PRF.PL
‘Then suddenly everybody hastened into the palace.’
d. Sil. ,
Subito vilis rubenti fluxit mulctra mero.
suddenly, worthless.NOM red.ABL flow.PFV.SG milk pail.NOM wine.ABL
‘Suddenly, the worthless milking pail flowed with red wine.’
e. Cic. Att. , ,
Se statim ad te navigaturum esse.
REFL.SG.ACC at_once at you.ACC sail.INF.FUT.ACC.M be.INF
‘That he was on the point of setting sail at once to join you’.
f. Cic. Fam. , , ,
Si statim navigas, nos Leucade consequere.
if at_once sail.SG us.ACC Leucas.ABL follow.FUT.SG
‘If you sail off at once, you will overtake me at Leucas.’
g. Lucr. ,
E terraque ex-orta repente arbusta
out earth.ABL=and out-rise.PTCP.PFV.NOM.N.PL suddenly tree(N)NOM.PL
salirent.
leap.IPFV.SBJV.PL
‘And trees would suddenly leap out of the turf.’
h. Enn. Ann. ,
Simul ex alto longe pulcherruma praepes
suddenly out high.ABL far beautiful.SUPERL.NOM nimble.NOM
laeva volavit avis.
left.NOM fly.PFV.SG bird.NOM
‘Suddenly there appeared to the left, in the distance, out of high heaven,
a most beautiful bird flying with good omen.’
() Latin; a sample of telic predicates headed by prefixed manner-of-motion verbs
a. Liv. , ,
Deinde subito ad arma dis-currerunt.
then suddenly at weapon.ACC.PL apart-run.PRF.PL
‘Then, suddenly, they ran in all directions for the weapons.’
Weak satellite-framed languages
root expressing the Manner component, FLU ‘flow’, identifying a co-event. The
s-framed pattern is preserved, since the Path and the verb correspond to different
phonological realizations.
The predicates in (), on the other hand, seem to be counterexamples to ().
Interestingly, however, out of the eight predicates headed by unprefixed verbs that
are shown in (), five are headed by curro ‘run’, salio ‘jump’, and volo ‘fly’. These
verbs exhibit a special behaviour in v-framed languages like Italian or French, a
behaviour which can be, despite appearances, compared with that of the Latin
cognate verbs of (). Specifically, these verbs can head constructions that at first
glance could be taken as CDMCs, which, as we know, are not possible in v-framed
languages. First, it has been observed that Italian correre ‘run’ and French courir ‘run’
may appear in predicates of bounded directed motion, as respectively illustrated in
()a and ()b. ()a additionally shows that bounded predicates with correre ‘run’
trigger selection of BE as auxiliary for the perfect tense and must, therefore, be
considered as heading an unaccusative predicate expressing a resultative event rather
than an activity:
() Italian and French; Folli and Ramchand (: ) and Pourcel and Kopecka
(: )
a. Gianni è corso in spiaggia in/*per un secondo.
John is run.PTCP.PFV.M.SG in beach in/for one second
‘John ran to the beach in a second/*for one second.’
b. Il court dans le jardin.
he runs in the garden
‘He runs into the garden.’
An analogous scenario can be described for verbs like jump and fly. Mateu (), for
instance, shows that these verbs display unaccusative behaviour in Italian—
specifically, BE-selection in the perfect—if accompanied with a PP:
() Italian; Mateu (: )
a. Gianni è/*ha volato a Roma.
Gianni is/*has flown to Rome
‘Gianni has flown to Rome.’
b. Gianni è/*ha saltato dalla finestra.
Gianni is/*has jumped from.the window
‘Gianni has jumped from the window.’
Crucially, not all verbs in Romance behave in this way. For instance, Folli and
Ramchand () show that Italian camminare ‘walk’ and galleggiare ‘float’, are
unable to license unaccusative predicates (with BE-selection) even in the presence
of a goal PP:
Weak satellite-framed languages
v PathP
Gianni Path’
Path PlaceP
pP PlaceP
in spiaggia Gianni Place’
Place corr
Latin and Slavic complex resultatives with an internal prefix
9
The analysis is inspired by a similar one in Mateu (). The difference between Mateu’s ()
analysis and mine is the treatment of the PP that typically appears in these Romance constructions of directed
motion. While I take it to be an adjunct to PlaceP, Mateu, assuming Late Insertion of roots, proposes that it is
the phonological realization of PlaceP (his Ploc projection), after the root has been incorporated into v (see
Haugen ). See Zubizarreta and Oh (), Den Dikken (), Gehrke (), Ramchand (), and
Real Puigdollers () for other analyses of similar phenomena in different languages.
Weak satellite-framed languages
v PathP
pP PathP
In this light, the cases of unprefixed predicates of directed motion with directional
DPs presented in section .., can be provided with the same analysis, on the basis of
the fact that the verbs licensing the directional DP (eo ‘go’, venio ‘come’) also show a
strongly directional character:
() Latin; Cic. Verr. Actio secunda, ,
Veniunt Syracusas.
come.PL Syracuse.ACC
‘They come to Syracuse.’
() Latin; analysis of ()
vP
v PathP
pP PathP
Path PlaceP
pro Place’
Place ven
In this analysis the verbal root VEN ‘come’ is merged as Compl-Place, identifying the
final location of the motion event, a deictic corresponding to the speaker. As is to be
expected, venio ‘come’, and also eo ‘go’, can also be found without a goal DP or PP:
() Latin; Caes. Gall. ,
Venisse tempus victoriae demonstrat.
come.INF.PRF time.NOM victory.GEN show.SG
‘He makes (them) see that the time of victory has come.’
Latin and Slavic complex resultatives with an internal prefix
10
See also Haverling (, a, b, , ), who claims that the prefixes provide bounds for
the event, and Romagno (, ), who considers them to be telicizers.
Weak satellite-framed languages
that it implies that all complex resultativity in Latin must involve prefixation. I also
do not support Van der Heyde’s (: ) claim that prefixes only attach to
‘determined’ verbs like venio ‘come’, highlighting the goal of motion inherent to
these verbs. It is evident from the discussion in this section and elsewhere in this
work, that preverbs may readily attach to verbs like blandior ‘flatter’ or equito ‘ride’,
which, in the absence of the preverb, do not imply a goal.
In sum, there is evidence that Slavic may indeed share with Latin the morpho-
logical requirement that I assumed in ().
The predicates above receive an analysis in which it is the root of the verb that is
merged as Compl-Place, identifying the final state of the event. Since, as we have
argued, this is also the position in which the prefixal root is first merged, a
natural explanation arises for the lack of obligatory prefixation in the above
predicates:
Gelatio Voice’
Voice vP
v PathP
arbores Path’
Path PlaceP
arbores Place’
Place nec
d. Vocabulary Insertion
nec-∅-∅-∅
In section .. the analysis in () was applied to telic directed motion constructions
based on verbs such as Latin curro ‘run’ () or venio ‘come’ () or Czech skočit
‘jump’ (). In this section I extend it to unprefixed change-of-location (COL)
alternants of the Locative Alternation and give-verbs.
In section ..., I proposed that COL alternants of the Locative Alternation
featuring an unprefixed verb like spargo ‘scatter’ involve an s-framed pattern, the
Weak satellite-framed languages
verbal root being merged as an adjunct to v and the directional PP being merged
directly as PlaceP:
() Cato, Agr.
Stercus columbinum spargere oportet
manure(N)ACC of_pigeon.ACC.N scatter.INF be_necessary.SG
in pratum.
in meadow.ACC
‘Pigeon manure must be scattered onto the meadow.’
() Analysis of ()
VoiceP
PRO Voice’
Voice vP
v PathP
Path PlaceP
stercus Place’
Place pratum
Place in
In an alternative analysis that explains the absence of the prefix in these constructions
the verbal root is first merged as Compl-Place and the directional PP is an adjunct to
either PathP (when the embedded DP is marked with accusative) or PlaceP (when
the embedded DP is marked with ablative):
() Analysis of ()
VoiceP
PRO Voice’
Voice vP
v PathP
pP PathP
p pratum stercus Path’
p in Path PlaceP
stercus Place’
Place sparg
Latin and Slavic complex resultatives with an internal prefix
The rationale for the above analysis, in which the verbal root occupies the position of
Compl-Place is the same as that adopted in the case of directed motion constructions
with unprefixed verbs like curro ‘run’ in section ... Roots like SPARG ‘scatter’ (and
also LIN ‘spread’ or STIP ‘cram’) are conceptually compatible with a syntactic-semantic
construal in terms of change of state of the thing scattered. A root like SPARG ‘scatter’
is expected, therefore, to be usable in the absence of any directional PP, as is the case:
By contrast, a strongly manner unprefixed verb like flo ‘blow’ does not license the
Locative Alternation, as shown in the entry for this verb provided by Gaffiot ()
and Lewis and Short (), since it is difficult to integrate its meaning as the final
state in a change-of-location event (the COL alternant). Its root FL ‘blow’ can,
however, be used as a Manner Co-event in COL alternants involving prefixed ad-flo
“at-blow” ‘blow upon’, which also license a COS alternant (Lemaire : ):
The fact that COL alternants involving the root FL ‘blow’ must bear a prefix and are
not licensed by a mere directional PP is indicative of the fact that the prefix is really
structuring the part of the configuration expressing the result.
Another interesting case of resultativity without internal prefixation is that of give-
verbs. As the following examples from Latin and Russian show, give licenses a telic
interpretation with a quantity object:
() Latin (Caes. Gall. , ) and Russian (Gehrke : )
a. Partem statim dederunt, partem [ . . . ] paucis diebus
part.ACC.SG at_once give.PFV.PL part.ACC.SG few.ABL.PL day.ABL.PL
sese daturos dixerunt.
REFL.SG.ACC give.INF.FUT.ACC.M.PL say.PFV.PL
‘These gave part of it at once, and said they would give the rest in a few
days.’
Weak satellite-framed languages
pro Voice’
Voice vP
v ApplP
ei Appl’
Appl PathP
Path PlaceP
Place d
A paraphrase for the above example would be ‘cause a tenth to go to his possession’.
In any case, it is clear why we do not find a prefix in predicates of pure transfer: the
Compl-Place position is already occupied by the verbal root. In turn, this analysis, in
which the dative is an ‘added argument’ (Pylkkänen ) easily explains why Latin
do ‘give’ is very frequently encountered without the dative, as shown in ()a. This
analysis of Latin do ‘give’ and its Slavic counterparts is compatible with the fact that
these verbs may sometimes appear with prefixes, as shown by non-transfer verbs like
Russian iz-dat’ “out-give”, ‘publish’ (cf. its exact Latin counterpart e-do):
Latin and Slavic complex resultatives with an internal prefix
The resulting combination of the prefix and the verb bears an idiosyncratic meaning,
‘publish’, which I take to be registered in the Encyclopaedia, in the entries for DA ‘give’
and IZ ‘out’, and triggered in the environment depicted in (), within the local
domain represented by vP.
In sum, the above data can be taken as supportive of the view that resultativity
depends, at least in change-of-state predicates, on the projection of specific structure
(PathP), and not on the presence of specific elements, the prefixes, which here are
assumed to contribute only conceptual content. It is only in the case of complex
resultativity that the presence of the prefix seems required, both in Latin and Slavic,
since the verbal root is associated directly to v and the position encoding the final
result/location can be occupied by the prefixal root. See section .. for an analysis of
yet another kind of unprefixed resultative predicate in Latin: AP resultatives based on
unprefixed verbs like ago ‘lead’ and facio ‘do’.
... Atelic predicates and prefixation The analysis proposed in this work, where
prefixation in Latin and Slavic is taken to result from an application of Raising of
Path to v, yields interesting predictions as to the shape of some atelic predicates with
a directional PP. Crucially, note that prefixation of Path to v is to be understood in
structural terms: v forms one word with the head of its sister PathP. Positing a Path-
to-v Raising operation aims at capturing this fact, since Raising, as an operation
previous to Vocabulary Insertion, is sensitive to structure, and not to linear adja-
cency. As a result, we expect no morphological relation to be effected between v and
any p head merged outside vP, that is, as an adjunct to vP, as shown in the following
diagram (where EA stands for external argument):
() VoiceP
EA Voice’
Voice vP
pP vP
Weak satellite-framed languages
11
The components for the search are as follows:
(i) Prefixed and unprefixed verbs (‘p-’ represents any prefix)
(p-)ambulo ‘walk’, (p-)curro ‘run’, (p-)equito ‘ride’, (p-)fluo ‘flow’, (p-)gredior ‘walk, step’, (p‑)labor
‘slip’, (p-)navigo ‘sail’, (p-)repo ‘crawl’, (p-)salio ‘jump’, (p-)volo ‘fly’
(ii) Atelicity-signalling expressions (cf. Vester : ff.; Torrego ; Pinkster : ff.; : ff.)
a. Adverbs
diu ‘for a long time’, diutule ‘for a little while’, paulisper ‘for a while’
b. Prepositions
per + quantified period of time ‘for’
c. Accusative forms of nouns and adjectives encoding periods of time
dies ‘day’, hora ‘hour’, nox ‘night’, mensis ‘month’, annus ‘year’, diurnus ‘of the day’, diutinus
‘lasting’, diuturnus ‘lasting’, nocturnus ‘of the night’, menstruus ‘which lasts a month’, menstrualis
Latin and Slavic complex resultatives with an internal prefix
containing each a motion verb, either prefixed or not, and an atelic adverbial
expression, ten sentences, shown in (), feature an unprefixed verb; in all these
sentences the durative expression is understood as temporally bounding the other-
wise unbounded (motion) activity expressed by the unprefixed verb:
() Latin examples with unprefixed manner-of-motion verbs and atelicity markers
a. Cels. ,
Tum diu ambulandum.
then for_long walk.PTCP.FUT.PASS.NOM.N.SG
‘Then one must walk for a long time.’
b. Plin. Epist. , ,
Diu iacui vel ambulavi.
for_long lie.PRF.SG or walk.PRF.SG
‘I have lain in bed or walked for a long time.’
c. Apul. Flor.
Ambulant diutule.
walk.PL for_a_while
‘They walk for a while.’
d. Ov. Am. , ,
Diu lacrimae fluxere per ora.
For_long tear.PL flow.PRF.PL through face.ACC
‘Tears flowed down her face for a long time.’
e. Liv. , ,
Nuntiatum est [ . . . ] sanguinis riuos
report.PTCP.PFV.NOM.N.SG is blood.GEN river.ACC.PL
per diem totum fluxisse.
through day.ACC whole.ACC flow.INF.PFV
‘It was reported that rivers of blood had flowed for a whole day.’
‘which lasts a month’, annuus ‘which lasts a year’, annalis ‘of a year’, annualis ‘a year old’,
aestas ‘summer’, hiems ‘winter’, ver ‘spring’, autumnus ‘autumn’, mane ‘morning’, vesper
‘evening’, vesperus ‘of the evening’, calendae/kalendae ‘calends’, idus ‘ides’, nonas ‘nones’,
lustrum ‘lustrum’, meridies ‘noon’, vigilia ‘time of keeping watch by night’, hibernus ‘of the
winter’, saeculum/seculum/saeclum ‘century’, saecularis ‘of a century’, aestivus ‘of the summer’,
aestivalis ‘of the summer’, vernus ‘of the spring’, vernalis ‘of the spring’, autumnus ‘of the
autumn’, autumnalis ‘of the autumn’, horalis ‘which lasts an hour’, matutinus ‘of the morning’,
postmeridianus ‘of the afternoon’, vespertinus ‘of the evening’, spatium ‘time span’
d. Adjectival suffixes indicating a period of time
-duus ‘of X days’, -ennius ‘of a year’, -noctius ‘of X nights’, -menstruus ‘of X months’, -menstris
‘of X months’
Weak satellite-framed languages
c. Mela ,
[Hypanis] [ . . . ] diu qualis natus est de-fluit.
Hypanis.NOM for_long how born.NOM.M.SG is down-flow.SG
‘The Hypanis flows down as it is in its spring for a long time.’
d. Mela, ,
Cyrus et Cambyses [ . . . ] [per] Hiberas et Hyrcanos
Cyrus.NOM and Cambyses.NOM through Iberian.ACC.PL and Hyrcanian.ACC.PL
diu [ . . . ] de-fluunt.
for_long down-flow.PL
‘The Cyrus and the Cambyses flow down through the lands of the Iberians and
Hyrcanians for a long time.’
In section ... we already saw cases like these, which are also pointed out by Van
der Heyde (: ) as problematic for an analysis of the Latin preverbs as
resultativity markers. I assume that these predicates display so-called fictive motion
(Jackendoff : ; Talmy ). They exploit the linguistic expression of motion
but they are interpretable as involving no motion at all:
() Talmy (: )
This fence goes from the plateau to the valley.
The pragmatically acceptable interpretation of () is one in which the fence is not
understood as undergoing a spatial transition from the plateau to the valley; rather, it
is understood to be as long as the space encompassed between the plateau and the
valley. As a result, the addition of an in-adverbial to () sounds odd, since it forces
the pragmatically aberrant interpretation whereby the fence is an entity actually
setting off from the plateau and arriving at the valley in a given amount of time:
() This fence goes from the plateau to the valley (#in an hour).
I understand that the Latin cases in () behave in the same way: the river, as a whole
entity, is not entailed to undergo a spatial transition. This licenses the atelic reading
signalled by the durative adverbials.
Another possible counterexample resulting from the search is ():
() Latin; Cat. Agr. ,
De-ambuletque horas IIII.
down-walk.SBJV.SG=and hour.ACC.PL four
‘He is to walk about for four hours.’
Note, first, that de- does not contribute here any spatial meaning, so the predicate is
not interpreted as ‘walk down’. Here I would like to suggest that de- is behaving as an
external prefix, licensing a quantification of the activity event, ‘to walk for an amount
of time’, much as does po- in the next Russian example:
Weak satellite-framed languages
12
These and other atelic predicates endowed with an internal prefix are problematic for Romagno
(, ), who establishes an explicit link between the telicizing effect of the preverb and its power of
supporting both transitive and unaccusative UOCs. In particular, telicization of the predicate by the
preverb requires the presence of an internal argument, which emerges either as direct object or, in the
case of unaccusatives, as sentential subject. See Romagno () for the same analysis applied to
preverbation in Ancient Greek.
Latin and Slavic complex resultatives with an internal prefix
cannot—measure the temporal extent of the motion event, but, on the contrary, the
temporal extent of the resulting state, in this case the state of having descended from
the chariots:
Finally, we find cases where the telic event encoded by the prefixed verb is interpreted
as iterated because of the presence of the durative adverbial. Thus, in ()a we are
forced to understand that an event of sailing off into the sea (e-navigat “out-sails”) is
repeated identically for some successive days (per aliquot dies continuos); in the
same way, the events of leaping at someone (ad-silient “at-jump”) or flying
around someone (circum-volant “around-fly”) of ()b and ()c, respectively,
are repeated ‘day and night’ (noctesque diesque, dies noctesque):
b. Stat. Theb. ,
Te volantes quinquaginta animae circum
you.ACC fly.PTCP.PRS.NOM.PL fifty spirit.NOM.PL around
noctesque diesque ad-silient.
night.ACC.PL=and day.ACC.PL=and at-leap.FUT.PL
‘The fifty spirits flying around you will leap at you day and night.’
c. Quint. Decl. ,
‘Dies’ inquit ‘noctesque miseranda patris
day.ACC.PL said night.ACC.PL=and pitiable.NOM.F father.GEN
umbra circum-volat’.
shadow(F)NOM around-fly.SG
‘ “Day and night,” he said, “does the father’s pitiable shadow fly around.” ’
Weak satellite-framed languages
In conclusion, there are reasons to think that the prediction made at the beginning of
this section is borne out for both Slavic and Latin: atelic predicates expressing a
directional but non-resultative, i.e. unbounded, motion are not internally prefixed.
Even if they sport some PP expressing the direction of the motion, since this PP is not
a sister to v, it cannot provide the material for the Raising-to-v operation to take
place, which gives rise to the prefixation effect.13
... A contrast between Latin and Slavic. The role of viewpoint aspect Notwith-
standing the unidirectional relation between prefixation and resultativity argued for
until now, it is only fair to acknowledge that telic predicates in Slavic languages
almost always bear an internal prefix, except those headed by a few verbs such as
Czech skočit ‘jump’ (see ()) or Russian dat’ ‘give’ (see ()b).14 There is controversy
about whether the simple imperfective forms of so-called incremental verbs like
Russian pisat’ ‘write’ or čitat’ ‘read’ may allow a telic interpretation, depending on
contextual factors, notably, when accompanied by an Incremental Theme. Berit
Gehrke (: , footnote ) contends that they may—see also Filip (:
). Crucially, however, the addition of the prefix cancels any atelic interpretation,
as pointed out by Gehrke (: ), in a way similar to the contrast involved in
pairs such as write/write down, read/read through and eat/eat up in English—see, for
instance, Borer (b: –). On the other hand, Krifka () and Borer
(b) champion the view that simple imperfectives in Slavic are always atelic.
The class of telic predicates in Slavic thus contrasts greatly with that of Latin,
where it is quite usual for an unprefixed verb still to head a telic predicate, as
illustrated in the following examples with capio ‘take’, facio ‘make’, neco ‘kill’, and
scribo ‘write’, which license the telic adverbial (in) paucis diebus ‘in a few days’:
() Latin; Bell. Afr. ,
Cirtamque oppidum [ . . . ] paucis diebus [ . . . ] capit.
Cirta.ACC=and town.ACC few.ABL.PL day.ABL.PL take.SG
‘And he conquers the town of Cirta in a few days.’
13
In this work I am not concerned with atelic, internally prefixed verbs displaying stative semantics,
like ab-sum “away-be” ‘be away’ or con-tineo “with-hold” ‘contain’. Assuming the theory of argument
structure configurations of section ..., these verbs, as states, involve a Place projection, which could
explain the prefixation if a vP-internal Place, as Path, were also marked as prefixal. Interestingly, Gibert
Sotelo (forthcoming), argues that these verbs are necessarily interpreted as stage-level and not individual-
level states (Carlson ; Kratzer ), which suggests that they may involve both a Path projection
encoding change and a stative BE verb encoding the state, as was also pointed out to me by Jaume Mateu
(p.c.). I leave this interesting hypothesis for future research.
14
See Filip (: ) and Bohnemeyer and Swift (: , footnote ) for more examples of
unprefixed telic verbs like Russian dat’ ‘give’, both in Russian and Czech.
Latin and Slavic complex resultatives with an internal prefix
15
See also Gehrke (: ) for the claim that internally prefixed predicates are telic and, by default,
perfective. An external prefix also marks perfectivity, but these prefixes carry an additional, quantificational
meaning of their own, such as ‘for a while’, ‘for a long time’, etc. (see section ..).
16
For the moment, I must remain agnostic about the actual implementation of the licensing of
perfective aspect by the prefixed verb. See Csirmaz () and É. Kiss (b) for an analysis of analogous
Weak satellite-framed languages
facts in Hungarian, a language in which there is a similar relation between perfectivity and telicizing
particles.
17
For the perfective/imperfective tense distinction in Latin, see Reinhold (); Vairel (); Pinkster
(: ff.); Haverling (), among others. For the interpretation of the perfect and its relation to
Aktionsart, and particularly to the conclusive/non-conclusive (in our terms, telic/atelic) distinction, see
Jensen (); Pinkster (: ).
The role of morphology in the analysis of resultative constructions
with the progressive. Importantly, then, the expression of perfectivity in Latin does
not have to rely on anything but dedicated morphology.18
b. Raising
[v HAMMER v] [Path [Place FLAT Place] Path]]
c. Linearization
HAMMER-v > FLAT-Place-Path
d. Vocabulary Insertion
hammer-∅ > flat-∅-∅
18
Needless to say, the morphological expression of the (im)perfective is obtained through different
morphophonological means, ranging from suffixal morphology to vowel changes, or even suppletion—see
Hewson () for a recent overview. What is important for my point, however, is that the expression of
(im)perfectivity is completely independent from internal (and external) prefixation.
Weak satellite-framed languages
b. Raising
[v [Path [Place EX Place] Path] [v TUSS v]]
c. Linearization
EX-Place-Path-TUSS-v
d. Vocabulary Insertion
ex-∅-∅-tuss-∅
Raising has the effect of forming a complex head with v, to which also the adjunct root
TUSS ‘cough’ raises, independently. In Latin and Slavic Raising affects also the Path
head, unlike in other s-framed languages like Germanic. The Vocabulary Item for Path
in these languages has the following shape (see section . for a examination of this):
The insertion frame of this Vocabulary Item ensures that Path must be linearly
preceded by Place and must linearly precede a span containing v. Since roots,
independently, raise to their dominating functional heads, the above Vocabulary
Item has the effect of bringing all the material up to the v head. If Place does not raise
to Path and Path does not raise to v, the contextual conditions for the insertion of the
exponent are not met and the structure fails to be interpreted. Thus, the univerbation
requirement in weak s-framed languages aims at accounting for the lack of AP
complex resultatives and PP complex resultatives headed by a simple (unprefixed)
verb in these languages. The univerbation of v, Path, and Place is incompatible with a
result predicate that is itself a word. I formulate this difference as a typological
hypothesis on the morphological properties of the Path head, as stated in ():
() The Split S-framedness Hypothesis
There are two types of s-framed languages: those with a morphologically
independent Path—strong s-framed languages—and those with an affixal
Path—weak s-framed languages.
The terms strong and weak are chosen to depict the fact that strong s-framed
languages are s-framed languages in a strong sense, in that the Core Schema,
expressed independently from the verb, may adopt any morphosyntactic form. On
the other hand, languages like Latin and Slavic are s-framed languages in a weak
sense, in that they pose morphological restrictions on the expression of the Core
Schema, much as it also is expressed as an element phonologically independent from
v in these languages. The terminology also aims at hinting at a diachronic develop-
ment in the morphosyntactic expression of complex transition events from Proto-
Indo-European down to Romance: strong s-framed Proto-Indo-European yielded
weak s-framed Latin, which yielded, in turn, v-framed Romance.
Next, I will show that the characterization of an s-framed language as weak, in terms
of the Split S-framedness Hypothesis, conspires with other independent morphological
factors of the language either to allow or to ban the formation of AP resultatives.
b. Croatian
Knjiga je plav-a.
book(F)NOM.SG is blue-NOM.F.SG
‘The book is blue.’
Following Mateu’s () and Kayne’s () proposal that adjectives and prepositions
involve the same basic category (see section ..), I take the (predicative) adjective to
be an instantiation of PlaceP. Specifically, in this case the head Place is endowed with
uninterpretable phi-features reflected in the agreement morphology, as shown in the
Latin and Croatian examples above. As described by Citko (: ff.), following
Chomsky () and Richards (), among others, uninterpretable features must be
deleted before arriving at the LF interface. However, since they may produce effects at
PF, as is the case in Latin and Slavic, they must also arrive at the phonological interface
first, which is implemented by positing that valuation of these features and transfer to
the interfaces is simultaneous. The PlaceP endowed with uninterpretable phi-features
(the adjective) is, thus, defined as a phase. This means, in turn, that such a PlaceP is not
computed within the cycle headed by v, given the theory of cyclic Spell-Out adopted in
section .. Let us see now what effect this scenario has on the derivation of complex
AP resultatives in Latin, such as that in the following made-up example:
() Latin; made-up ungrammatical example
*Ovidia poculum vacu-um bibit.
Ovidia.NOM goblet(N)ACC.SG empty-ACC.N.SG drink.SG
() PF-derivation of ()
a. Structure delivered by syntax
[VoiceP [Voice’ Voice [vP [v v BIB] [PathP [Path’ Path {[PlaceP [Place’ Place
VACU]]}]]]]]
b. Raising
[v Path [v BIB v]]
c. Linearization
Path-BIB-v
d. Vocabulary Insertion
?-bib-∅
At Vocabulary Insertion, after Raising and Linearization have taken place, Path
linearly precedes the root and the verb. However, Place does not precede it, since it
has been spelled out in a different cycle. Since the insertion frame in the Vocabulary
Item of Path specifies this condition, namely that Place should precede Path, the
exponent cannot be inserted, and the derivation fails:
() Path ⟷ ∅ / Place-_-[ . . . v . . . ]
The role of morphology in the analysis of resultative constructions
In section . I will examine the structure of this Vocabulary Item and I will try to
derive the same results on more principled grounds. For the time being, though,
consider that this Vocabulary Item would be compatible with a derivation in which
the adjective is an adjunct to PlaceP, which contains a prefixal root at Compl-Place.
Such an analysis could perhaps be applied to cases like the following, described by
Pinkster (: –), and other cases shown already in section ..:
() Verg. A. ., in Pinkster (: )
Tum sterilis ex-urere Sirius agros.
then sterile.ACC.PL out-burn.INF Sirius fields.ACC
‘Then did Sirius burn the fields sterile.’
() Analysis of ()
VoiceP
Sirius Voice’
Voice vP
v PathP
v ur agros Path’
Path PlaceP
{pP} PlaceP
p steril agros Place’
Place ex
In this case, the adjective does not correspond to the PlaceP of the spinal derivation,
since it is an adjunct. It is spelled out too in a different cycle from that headed by v.
Since the PlaceP containing the prefixal root EX ‘out’ does not constitute an inde-
pendent cycle, Raising can take place from this position and the insertion frame of
the Vocabulary Item for Path is satisfied. Although such cases are fully compatible
with the analysis developed here, Pinkster (: ) points out their scarcity, which
makes them suspect. In particular for the case in hand, sterilis ‘sterile’ could be an
AP modifying the object agros ‘fields’ directly. See section .. for more related
discussion.
In the next section I will show how Latin allows simple (i.e. non-complex)
adjectival resultative constructions. It remains also to be seen whether within the
class of s-framed languages there can be prefixation of a resultative adjective when
the adjective is not inflected. This is what we find in Icelandic and Mandarin Chinese,
as will be shown in section .. and ..
Weak satellite-framed languages
Observe that the conditions for insertion of the above Vocabulary Item pre-empt the
insertion of the other null exponent of this functional head:
The role of morphology in the analysis of resultative constructions
PRO Voice’
Voice vP
v PathP
servitutem Path’
Path {PlaceP}
servitutem Place’
Place len
b. Raising
[Voice [v Path v] Voice]
c. Linearization
Path-v-Voice
d. Vocabulary Insertion
red-d-∅
The exponent of v in these cases is also predictable, like the light verb do ‘give’ or
linquo ‘leave’, and it is triggered directly by re. The contextual specification of the
Vocabulary Item for Path in these constructions has the power to preclude another
pattern of complex AP resultatives which is not found in Latin:
19
The analysis is inspired by that proposed by Folli and Harley () for the so-called faire par
causative constructions (Kayne ) in Italian. On the basis of different syntactic and semantic facts these
authors propose that causative constructions with a causee introduced by da ‘by’, and not by a ‘to’, in
Italian, involve a nominalized VP (riparare la macchina ‘repair the car’, below) and a creation, rather than
causative, fare ‘do’ (Folli and Harley : ):
(i) Gianni ha fatto riparare la macchina a/da Mario.
Gianni has made repair.INF the car to/by Mario
‘Gianni got Mario to repair the car.’ / ‘Gianni got the car repaired by Mario.’
The role of morphology in the analysis of resultative constructions
pro Voice’
Voice vP
v nP
n PlaceP
senatum Place’
Place ferm
The predicative relation between the object and the adjective is articulated through a
Place head. I hypothesize that PlaceP is taken as complement by a null n head, which
nominalizes it. Importantly, since the above structure does not involve the projection
of PathP, no prefix is expected, under the assumption that prefixation in Latin (and
Slavic) involves Raising of Path onto v.
There is independent evidence that in Latin a Small Clause combining a DP
subject and a participle, AP, or DP predicate can function wholly as an argument,
which reveals that it is in fact nominalized. This is shown by the following examples
from Pinkster (: ), in which I have italicized the nominalized Small Clause,
which in both cases is acting as the sentential subject:
In the above examples the thematic relation is established between the main predi-
cate (augebat metum ‘increased the fear’, ostenderet ‘would show’) and the whole
italicized sequences, and not between the main predicate and the head of those
sequences (hostis ‘the enemy’, filius legati ‘their commander’s son’). This is evidence
that those italicized sequences, which contain, as I have said, a predication relation,
are functioning as fully fledged arguments.
. Summary
In this chapter I have located Latin within the wider cross-linguistic scenario with
respect to the way it syntactically builds complex events of change. Departing from
the results arrived at in Chapter , that is, that Latin qualifies as an s-framed
language, I have shown that, nevertheless, it differs from other s-framed languages
such as the Germanic languages in disallowing typically s-framed constructions
based on AP resultative predicates. Since neither Talmy’s typology nor the subse-
quent revisions thereof predict such a scenario, I have tried to seek out a possible
explanation for this behaviour. My first step has been to observe that Latin patterns in
this sense with the group of Slavic languages, also acknowledged for their s-framed
status, and I have focused on an additional feature that characterizes both: the fact—
arrived at for Latin through corpus searches—that complex resultative constructions
are always built on prefixed verbs. I have proposed that in these languages there is a
general univerbation requirement bringing Place and Path to form a complex head
with v. This requirement, encoded in the insertion frame of the head Path, is
incompatible with the situation in which the PlaceP is an adjective inflected for
agreement, since it forms an independent Spell-Out cycle. It is only when Path is
strictly left adjacent to v, in simple resultative constructions, that Path receives an
exponent (re) independently of the realization of PlaceP, which can thus be realized
as an adjective.
6
In this chapter I explore the empirical coverage of the Split S-framedness Hypothesis,
both for strong and weak s-framed languages. Icelandic will be shown to provide data
underpinning the assumption that the morphological characterization of Path and
the inflectional morphology on the resultative adjective are the factors at stake in
triggering the split within the s-framed class of languages. As stated in the Split
S-framedness Hypothesis (section ..), there are two basic types of s-framed
languages: those where Path can be morphologically independent—strong s-framed
languages, and those where Path is always an affix and must then lean on another
head to be licensed—weak s-framed languages. If v-framed languages are taken into
account, a three-way typology emerges based on the phonological dependence of
Path with respect to the verb. At one extreme are strong s-framed languages, where
the Path is morphologically independent from the verb, both being expressed as
different morphemes and words. These languages allow the generation of PP,
particle, and AP resultatives. Next to these languages are weak s-framed languages,
in which the Path and the verb are different morphemes but one phonological word.
This allows resultatives based on affixal particles but precludes the formation of PP
resultatives and of AP resultatives if the predicative adjective is inflected. Finally,
I compare previous analyses of the issue of the expression of resultativity and change
of location cross-linguistically, and address particular problems.
Unfortunately, Horrocks and Stavrou () do not specify whether the search they
performed included the prefixed counterparts of ‘verbs meaning “walk”, “run”,
“swim”, and “sail”’. Geoffrey Horrocks, in a personal communication, informs me
that, in fact, the search was carried out taking into account only unprefixed verbs.
According to my assumptions and the hypothesis that Ancient Greek is a weak
s-framed language, the results of Horrocks and Stavrou’s () search are unsurpris-
ing: unprefixed verbs in weak s-framed languages cannot support telic complex
resultative constructions, even if accompanied by an alleged goal PP.1
However, prefixed predicates are telic in Ancient Greek, even in the absence of
directional PPs, in conformity with present assumptions about weak s-framed
languages. Thus, in performing a search of prefixed motion verbs in a subcorpus
of the Thesaurus Linguae Graecae (Pantelia ), a non-trivial amount of unam-
biguously telic examples were found:2
() Ancient Greek; telic manner-of-motion predicates with a prefixed verb
a. Thucydides, Historiae, , , ,
Kéra:i tôn Athe:naío:n
wing(N)DAT.SG the.GEN.M.PL Athenian(M)GEN.PL
euthùs apo-bebe:kóti.
right_away away-step.PTCP.PFV.DAT.N.SG
‘The wing of the Athenians, which had just disembarked’.
1
See also Lavidas (: ), who acknowledges that the prefixes in Ancient Greek can mark ‘the
completion of the action denoted by the verb’.
2
Verbs searched for: apo-baíno: ‘walk, step away’, ek-baíno: ‘walk, step in’, em-baíno: ‘walk, step in’,
kata-baíno: ‘walk, step down’, án-eimi ‘go up’, áp-eimi ‘go away’, eís-eimi ‘go in’, kát-eimi ‘go down’, ap-
hippeúo: ‘ride away’, kat-hippeúo: ‘ride down, over’, ana-kolumbáo: ‘come up after diving’, apo-kolumbáo: ‘dive
and swim away’, eis-kolumbáo: ‘swim into’, ek-kolumbáo: ‘swim ashore, plunge into the sea from’, kata-
kolumbáo: ‘dive down’, ana-pléo: ‘sail upwards, go up-stream, rise to the surface’, apo-pléo: ‘sail away’, eis-pléo:
‘sail into a harbour’, ek-pléo: ‘sail out’, kata-pléo: ‘sail down, back’, ana-trékho: ‘run back’, apo-trékho: ‘run off,
away’, eis-trékho: ‘run in’, ek-trékho: ‘run out’, en-trékho: ‘run in, enter’, kata-trékho: ‘run down’. The subcorpus
of authors consisted of non-late (pre-Christian) authors (and corpora): Aeschylus, Alcidamas, Anonymi
medici, Antiphon, Pseudo-Apollodorus, Aristophanes, Aristotle and the Corpus Aristotelicum, Bacchylides,
Chariton, Demosthenes, Epicurus, Euclid, Euripides, Hesiod, Homer, Isocrates, Lysias, Plato, Plutarch,
Sophocles, Thucydides, Vettius Valens, Xenophon, and the Scholia in Aeschylum. The references for examples
provided here are those provided by the Thesaurus Linguae Graecae (Pantelia ). The transliterations of all
the Ancient Greek examples in this section are my own. I am grateful to Geoffrey Horrocks for suggesting the
kind of adverbial or case marked DP I should use as the telicity-signalling expression in Greek.
A revision of Talmy’s typology
i. Thucydides, Historiae, , , ,
Kaì hoi Athe:naîoi ou pollô:i
and the.NOM.M.PL Athenian(M)NOM.SG not much.DAT.N.SG
hústeron kata-pleúsantes.
later.ACC.N.SG down-sail.PTCP.AOR.NOM.M.PL
‘The Athenians having sailed up to there not much later’.
j. Thucydides, Historiae, , , ,
Ek-dramóntes áphno: ek tê:s póleo:s.
out-run.PTCP.AOR.NOM.M.PL suddenly out the.GEN.F.SG city(F)GEN.SG
‘Having run out of the city all of a sudden.’
k. Xenophon, Hellenica, , ,
hoi mèn psiloì euthùs ek-dramóntes
the.NOM.M.PL PART light(M)NOM.PL right_away out-run.PTCP.AOR.NOM.M.PL
e:kóntizon.
hurl_javelins.IPFV.PL
‘The light troops, having run out immediately, started hurling javelins.’
Note, importantly, that in the examples above the directional PP or the DP is
optional—see ()h and ()j for cases of the former and ()c for a case of the latter.
If in Ancient Greek, as the data seem to suggest, complex resultatives feature a prefix
representing Path, it should count as a weak s-framed language, within present
assumptions. Since the predicative adjectives in Ancient Greek are always inflected
for agreement, as shown below, the prediction emerges that this language will not
allow complex adjectival resultative constructions.
() Ancient Greek; Pseudo-Apollodorus, Bibliotheca (sub nomine Apollodori), ,
Toûto dè hupermégeth-és estin.
this.NOM.N.SG PART exceedingly_difficult-NOM.N.SG be.SG
‘This is exceedingly difficult.’
As far as my (limited) competence in Ancient Greek tells me, those constructions are
not found in Ancient Greek. This is also hinted at by Horrocks (: ) and, most
importantly, it is claimed as an empirical fact by Horrocks and Stavrou (: ),
who point out that a search ‘for predicate adjectives in the Thesaurus Linguae
Graecae electronic database of ancient Greek literature (www.tlg.uci.edu) produced
no examples of result-state readings, and only depictive ones’.
Ancient Greek thus turns out to be a weak s-framed language, since, being s-framed
(recall the UOCs in () above), it does not seem to accept a morphologically inde-
pendent Path, as suggested by an analysis of CDMCs in this language. As a result,
Ancient Greek is correctly predicted not to license adjectival resultative constructions.
A revision of Talmy’s typology
Similarly, the Dutch particles in ‘in’ and af ‘off ’ describe the final location of the truck
and of the hat in ()a, and ()b, respectively, and in ‘in’ describes a more abstract
location or state (the state of being available) for the petition in ()c (crucially, in in
the first example must be understood as a particle, and not a postposition—see Van
Riemsdijk : ):
McIntyre points out that ein in these predicates is not a mere inchoative marker
whereby ein-spielen “in-play” for instance, should be rendered ‘begin playing’.
Rather, these predicates present an interpretation analogous to adjectival resultative
constructions such as the following:
Dutch particles also display the power of introducing an argument, as af ‘off ’ and toe
‘to’ illustrate in the following examples:
Finally, German and Dutch particles may induce telicity in the predicate in which they
appear, which suggests, within the framework adopted here, the projection of PathP. For
instance, the atelic process portrayed by schmort ‘braises’ in ()a is turned into an
accomplishment (where the result state of the referent of the object is that of being partially
or lightly affected by the action—see McIntyre ) in ()b by virtue of the particle an ‘on’:
Analogously, Levin and Rappaport Hovav (: –) report that Dutch atelic
bloeien ‘bloom’ (see ()a) turns into a telic change-of-state predicate with the
addition of op ‘up’ (see ()b):
As can already be gathered from the examples above, particles in these languages
may appear attached to the verb or separated from it. The following minimal pair in
German illustrates both possibilities:
() German; Zeller (b: )
a. Weil Peter in den Bus ein-steigt.
because Peter in the.ACC bus in-climbs
‘Because Peter gets on the bus’.
b. Peter steigt in den Bus ein.
Peter climbs in the.ACC bus ein
‘Peter gets on the bus.’
According to Zeller (b) and Lüdeling (), in subordinate clauses like
()a the particle appears adjacent to the verb, which sits in its original final
position. In matrix clauses, however, the verb undergoes movement to a ‘second’
position (the well-known phenomenon of V—see Haider and Prinzhorn ();
Weerman (); Vikner (), among others), stranding the particle. Dutch
particles are also separable from the verb through stranding under V
movement:
() Dutch; Booij (: )
a. . . . Hans zijn moeder op-belde.
Hans his mother up-called
b. Hans belde zijn moeder op.
Hans called his mother up
c. . . . de fietser neer-stortte.
the cyclist down-fell
d. De fietser stortte neer.
the cyclist fell down
A second proof of the morphological independence of the particle is the fact that it
may be fronted under topicalization:
.. English
As for Dutch and German, following a long-standing tradition in which particles are
merged as part of a Small Clause (Stowell ; Kayne ; Hoekstra ; Den
Dikken ; Svenonius ; and Hale and Keyser , among others), I assume
that English particles also signal the projection of a PathP structure, and that the
particle originates as a root in Compl-Place position. With that in mind, we can quite
safely claim that Path is not obligatorily prefixed onto the verb in this language, as the
following examples show:
() Svenonius (: )
a. The doorman threw the drunks out.
b. The firefighters hoisted the equipment up.
c. The police chased the demonstrators off.
This state of affairs is in accordance, under present assumptions, with the well-
known allowance of adjectival resultatives in this language:
A revision of Talmy’s typology
There is evidence that we are dealing with two outs here. On the one hand, the
semantics are clearly different, and, most notably, the prefixed out never delivers an
idiomatic meaning, such as the one we find in (). On the other hand, the
phonologies are also different, since, though segmentally identical, the out in ()
is a prefix, and that in () cannot be prefixed. I propose then, that out-prefixation
depends on the idiosyncratic properties of this out root and not on the properties of
Path in English, which, as discussed, is not required to be prefixed. Thus, as in the
case of verbal prefixes in Dutch and German seen in the previous section, the out- of
out-prefixation has the following Vocabulary Item:
This analysis can be applied, on the other hand, to other cases of prefixes in
predicates arguably involving a PathP. Marantz (), for instance, proposes that
in Latinate verbs like construct and destroy the segments con- and de- are actually
Strong s-framed languages
predicates of the internal argument and end up prefixed onto the verb. Harley ()
applies the same analysis to Latinate verbs in general (like com-pose, dis-sect, ex-hibit,
in-cise), accounting for the failure of these complex verbs to combine with particles,
with resultative predicates or to head double object constructions.3
.. Icelandic
Icelandic resultative constructions may feature non-prefixed particles:
() Icelandic; Den Dikken (: )
a. Ég gaf (*upp) Maríu (upp) símanúmerið mitt (*upp).
I gave up Maríu up phone_number my up
‘I gave Mary my phone number.’
b. Í gær hafa þeir sent (*upp) strákunum (?upp) peningana (upp).
yesterday have they sent up boy.the.PL up money.the up.
‘Yesterday the boys sent up the money.’
c. Ég hef rétt (*niður) Jóni (?niður) hamarinn (niður).
I have passed down John down hammer down
‘I have passed John the hammer.’
Moreover, resultatives may be licensed singly by a PP, without the aid of either
particle or prefix:
() Icelandic; Whelpton (: )
Báturinn flýtur undir brúna.
boat.the floats under bridge.the.ACC
‘The boat is going under the bridge floating.’
We expect, accordingly, that adjectival resultatives are allowed in Icelandic, as is
the case:
() Icelandic; Whelpton (: –)
a. Járnsmiðurinn barði málminn flatan.
blacksmith.the pounded metal(M)the.ACC.SG flat.ACC.M.SG
‘The blacksmith pounded the metal flat.’
b. Ég kýldi hann kaldann.
I punched him.ACC.SG cold.ACC.M.SG
‘I punched him out cold.’
3
As for prefix re- (rewrite, reopen, reconsider, etc.), which induces the presupposition that the state
codified by the predicate had previously existed, I will assume that it is an adjunct merged lower than v,
modifying PlaceP. See Marchand (); Keyser and Roeper (); Lehrer (); Lieber (); Marantz
(, ); and Acedo-Matellán () for discussion on the syntax and semantics of re-.
A revision of Talmy’s typology
c. Að nudda þá slétta.
to rub them.ACC.M.PL smooth.ACC.M.PL
‘To rub them smooth’.
d. [Þá] slengdi illi andinn honum flötum.
then slung evil spirit.the him.DAT.M.SG flat.DAT.M.SG
‘Then the evil spirit slung him down flat.’
e. Þeir dældu hana fulla af lyfjum.
they pumped her.ACC.F.SG full.ACC.F.SG of drugs
‘They pumped her full of drugs.’
f. Dóra æpti sig hás-a.
Dóra screamed herself.ACC.SG hoarse.ACC.F.S
‘Dóra screamed herself hoarse.’
g. [Hann] reif hurðina opna.
he tore door(F)the.ACC.SG open.ACC.F.SG
‘He tore the door open.’
h. Hann skrúbbaði pönnurnar hreinar.
he scrubbed pot(F)the.ACC.PL clean.ACC.F.PL
‘He scrubbed the pots clean.’
i. Þvo mig hreinan.
wash me.ACC.M.SG clean.ACC.M.SG
‘Wash me clean.’
However, what is most interesting about Icelandic for the present discussion is that it
presents two types of adjectival resultative constructions. We encounter those in
which the adjective is morphologically independent from the verb, as in the examples
in () above, and those in which it is prefixed to the verb, as the following examples
show:
() Icelandic, Whelpton (: , : )
a. Hann hvít-bæsti rammann.
he white-stained frame.the
‘He stained the frame white.’
b. Svart-litaður.
black-coloured.NOM.M.SG
c. Þunn-sneiddu sveppirnir.
thin-cut.NOM.M.PL mushroom(M)the.NOM.PL
‘Thin-cut mushrooms’.
d. Fín-muldu piparkornin.
fine-ground.NOM.N.PL peppercorn(N)the.NOM.PL
‘Fine-ground peppercorns’.
Strong s-framed languages
e. Hrein-skrúbbuðu pönnurnar.
clean-scrubbed.NOM.F.PL pot(F)the.NOM.PL
‘Clean-scrubbed pots’.
f. Mjúk-brædda súkkulaði.
soft-melted.NOM.N.SG chocolate(N)NOM.SG
‘Soft-melted chocolate’.
Crucially, the data in () show a correlation between prefixation of the adjective and
lack of agreement morphology, while the data in () show that when the adjective is
inflected, it is not prefixed. This could be interpreted in the following way: the
adjective in Icelandic (complex) resultative constructions is allowed to bear agree-
ment morphology. If it does, it cannot be attached to the verb, but this does not yield
a deviant ouput, since the Path is not specified as affixal in this language. When it
does not bear agreement morphology, however, it may be prefixed to the verb.4 Thus,
Icelandic subsumes two logical types of strong s-framed languages: those where the
adjective is inflected and those where it is not inflected. As one might expect, it is only
in the latter that adjectival resultatives of the English type are permitted. In this sense,
Icelandic lends support to the hypothesis that the allowance of this type of resultative
constructions depends, first, on the morphological features of Path and, second, on
the requirement that predicative adjectives bear agreement morphology.
.. Finno-Ugric
Outside Indo-European (at least) two Finno-Ugric languages are found that pattern
with Germanic in being strong s-framed languages: Finnish and Hungarian.
In Finnish the verb does not seem to require the appearance of a Path-signalling
affix in resultative constructions based on PPs or particles:
() Finnish; Fong (: ) and Kolehmainen (: )
a. Toini tanssi huonee-seen / huonee-sta.
Toini danced room-ILL room-ELA
‘Toini danced into/out of the room.’
4
Both Dutch and German feature particle verb constructions where the particle is an adjective and
appears adjacent to the verb in verb final environments. I illustrate this with Dutch:
(i) Dutch; Booij (: )
a. . . . Jan het huis schoon-makte.
Jan the house clean-made
b. Jan maakte het huis schoon.
Jan made the house clean
However, since in these languages the adjective is only inflected when used attributively, and not
predicatively (consider, for instance, Ger. Das weiss-e Buch, ‘The.NOM.N.SG white-NOM.N.SG book(N)NOM.SG’
vs Das Buch ist weiss(*e)), the formal dissociation shown by Icelandic adjectival resultatives does not
obtain.
A revision of Talmy’s typology
As for Hungarian, this language possesses a set of particle-like elements that are
readily amenable to an analysis in terms of resultative particles analogous to those we
have described for other languages—see Perrot (: ) and É. Kiss (, a).
That these elements are good candidates of Path(P) is the fact that they describe the
final state of a motion event (see ()), affect the telicity of the predicate (see the
diagnostics with temporal modifiers in ()), and may introduce unselected objects
(see ()):
Strong s-framed languages
() Hungarian; Horvath (), in Julien (: ) and Hegedűs (: )
a. János ki-ment.
János out-went
‘János went out.’
b. János át-jött.
János over-came
‘János came over.’
() Hungarian; É. Kiss (: –)
a. János hétfőre *(el) olvasta a regényt.
János by_Monday PART read.PST the novel
‘János read the novel by Monday.’
b. János egész este (*el) olvasta a regényt.
János whole evening PART read.PST the novel
‘János read the novel the whole evening.’
() Hungarian; Bende-Farkas ()
A kutya *(fel-)ugatta a szomszédokat
the dog up-bark.PST the neighbour.ACC.PL
‘The dog woke the neighbours with its barking.’
Hungarian particles are not obligatorily affixed to the verb. It is true that, as shown by
É. Kiss (: ), when there is no logical operator in the sentence the particle must
form one and the same phonological word with the verb:
() Hungarian; É. Kiss (: )
János [ω fel olvasta] a verseit.
János up read.PST the poems
‘János read out his poems.’
However, there is a variety of syntactic conditions that may disrupt the morpho-
logical connection between the particle and the verb: the presence of negation (see
()a), contrastive topicalization of the particle (see ()b), or even movement into a
matrix clause (see ()c and ()d) (see also Farkas and Sadock ; Puskás :
ff.; and É. Kiss a):
() Hungarian; É. Kiss (: –)
a. Péter nem olvasta őket fel.
Péter NEG read.PST them up
‘Péter did not read them out.’
b. Fel csak János olvasta a verseit.
out only János read.PST the poems
‘Only John read his poems out loud.’
A revision of Talmy’s typology
() Hungarian
a. Snyder (: )
A munkás lapos-ra kalapácsolta a fémet.
the worker flat-TRANSL hammer.PST the metal
‘The worker hammered the metal flat.’
b. Bende-Farkas (: )
Mari beteg-re ette magat.
Mari sick-onto eat.PST.SG self.ACC
‘Mari ate herself sick.’
c. Csirmaz (: )
János tisztá-ra mosta a ruhát.
János clean-onto washed the dress
‘János washed the dress clean.’
Note, finally, that both in Finnish and in Hungarian the adjective heading the
resultative predicate is marked with a special case: translative -ksi in Finnish (see
()) (Levinson ) and sublative -ra/-re in Hungarian in () (Marácz ). That
this case mark signals resultativity is shown by the fact that in Finnish depictive
secondary predication, unlike resultative secondary predication, requires the essive
case (-na):
This morphological fact fits nicely with the analysis put forward here in which the
resultative adjective, encoding final state, is embedded within a PathP: the translative
case of Hungarian and Finnish would correspond to the Vocabulary Item for the
Path head, to which Place raises—see section .. for the same analysis of AP
resultative constructions in English. I illustrate this with the PF-derivation of the
Finnish complex AP resultative of ()a:
Mari Voice’
Voice vP
v PathP
v jo teekannun Path’
Path PlaceP
teekannun Place’
Place
b. Raising
[v JO v] [Path [Place TYHJÄ Place] Path]
c. Linearization
JO-v > TYHJÄ-Place-Path
d. Vocabulary Insertion
jo-∅ > tyhjä-∅-ksi
Strong s-framed languages, with a non-affixal Path, do not require (although they
allow) the univerbation of v and PathP, so these two elements may be realized
independently. As a result, constructions in which the v head is associated with a
root encoding a Co-event are predicted as being possible, since this association does
not interfere with the morphological realization of the material in PathP.
In weak s-framed languages Path has generally a null exponent and must be affixed
to a span of nodes containing the verb and the rest of the material in PathP (see
section ..). This has the effect of ruling out structures in which the material in
PathP may be realized independently, such as adjectival resultative constructions. It
is only when Path is strictly adjacent to v, that is, in simple resultative constructions,
that, although prefixed onto v, it receives an exponent independently of the realiza-
tion of PlaceP, which can thus be realized as an adjective (see section ..).
Finally, in v-framed languages, Path has a null exponent and must be strictly adjacent
to v. This has the effect of precluding the presence of any Co-event root associated with v,
since such a root would be linearized between Path and v. It also has the effect of yielding
the effect that Path and v are always realized as the same morph. V-framed languages
admit constructions with PPs and APs encoding a resulting location or state provided
that the above-mentioned condition on the realization of Path is met. Thus, locative PPs
with simple verbs like Cat. anar ‘go’ and resultative APs with simple causative verbs like
Cat. deixar ‘leave’ are fine in these languages (see section ...).
Importantly, the properties of the Path head may interfere with independent properties
of the language to yield either allowance or disallowance of complex resultative construc-
tions based on APs. It has been argued that in Latin and Slavic the fact that adjectives are
inflected for agreement makes them constitute independent Spell-Out cycles. Thus, when
an AP is merged as PlaceP in a transition predicate, the univerbation requirement pointed
out above, and encoded in the Vocabulary Item of Path, cannot be met. Indirect evidence
of the role of agreement inflection in the adjective in licensing or not adjectival resultative
constructions comes from the strong s-framed language Icelandic, since, although the
adjective is allowed to be prefixed onto the verb, this happens only when it is not inflected.
A typology based on the morphology of Path. The case of Mandarin
5
See Talmy (), Peyraube (), Xu (), Mateu (), and Fan () for arguments that
Chinese is an s-framed language. For other divergent analyses see Tai (), Slobin (, ), and
Chen and Guo ().
A revision of Talmy’s typology
if, as discussed in Acedo-Matellán and Mateu (), Path starts out as an inde-
pendent element, then it optionally attaches onto the verb, afterwards the affixation is
obligatory, only allowing the stranding of PlaceP when Path is strictly adjacent to the
verb, and, lastly, it becomes phonologically indistinguishable from the verb, this last
option modelled here as emerging from a requirement of strict adjacency between
Path and v.6
6
See Eythórsson () and Acedo-Matellán and Mateu (, ) for similar considerations, and
Verkerk () for discussion of the status of proto-Indo-European as an s-framed language on the basis of
a phylogenetic comparative study. See also Coleman (), Untermann (), Vincent (), and
Oniga () for other theories of the emergence of preverbs in Latin. See Stolova () for a lexical study
of Late Latin as an intermediate stage between s-framed Latin and v-framed Romance.
Previous approaches and possible counterexamples
event. The perfective form of a telic verb, on the other hand, presents a final temporal
bound which coincides with the telos inherent in the lexical semantics of the verb. As
regards imperfective forms, they imply viewing the event without bounds. But whereas
imperfective atelic verbs convey an event unfolding continuously or randomly, imper-
fective telic verbs entail an incrementality towards a goal which, crucially, is not
implied to be attained. Consequently, for Horrocks and Stavrou () inner aspect
has to be determined before grammatical aspect, in order to compute an overall
aspectual value for the verb. But, since, in languages like Ancient or Modern Greek,
either the perfective or imperfective form or stem—in the many cases of suppletion—
has to be chosen before inserting the verb into the tree (given that quite often one form
is not predictable from the other), it follows that the inner-aspectual value must have
also been determined before that insertion, and cannot interact with the syntactic
environment of the predicate. This is the reason why in languages where that mor-
phological/lexical choice is forced, the semantics of a non-terminative verbal lexeme
cannot interact with syntactic material, such as a goal PP or a resultative adjective, to be
rendered terminative. By contrast, in languages lacking such grammaticalized perfect-
ive/imperfective opposition the overall aspectual value of the verb is not fixed when it is
inserted in the tree and can, therefore, interact with the syntax in constructions such as
adjectival resultative constructions and CDMCs.
I detect two incompatibilities between these accounts and my own. The first
concerns the existence of languages with a grammaticalized opposition between
perfective and imperfective that do license, however, complex resultative construc-
tions, pace Horrocks and Stavrou (, ) and Horrocks (): Latin, Slavic,
and Ancient Greek. The fact that in these languages those complex resultative
constructions are always based on prefixed particles does not make them less
complex resultative constructions, with a resulting state/location encoded by the
prefix and a differentiated event leading to it encoded by the verb.
The second problem is restricted to the accounts in Horrocks and Stavrou ()
and Horrocks (), and not to that in Horrocks and Stavrou (). In these analyses
a dissociation is made between the availability of adjectival resultative constructions
and that of CDMCs. While the availability of the former depends, as we have seen, on
the absence of a grammaticalized perfective/imperfective opposition, the availability of
the latter is subject to the fact that the language in question possesses the formal means
to unambiguously express goals (that is, telic Paths) in PPs.7 In particular, Ancient
Greek and English are shown to be able to express bounded Paths with dedicated
prepositions (English to, Ancient Greek eis ‘into’) and, in the case of Ancient Greek,
(accusative) case. Thus, in Ancient Greek predicates headed by a manner-of-motion
verb and accompanied by a goal-encoding PP, the verbs are claimed to be reclassified
7
An analysis based on the lexical availability of particular prepositions is also that adopted by Folli and
Ramchand (, ), Son (), and Son and Svenonius ().
Previous approaches and possible counterexamples
trinken
[cause]
Crucially for the position expounded in this chapter and in Chapter , Kratzer also
appeals to inflectional morphology on the predicative adjective as a factor regulating
the licensing of adjectival resultatives. However, her use of this factor is different
A revision of Talmy’s typology
from mine: she contends that languages in which the predicative adjective obliga-
torily bears inflectional morphology cannot license adjectival resultatives because, for
an adjective to function as resultative, the null CAUSE morpheme must be affixed
onto it, a factor that precludes further affixation of the inflectional morphology.8
However, she herself already notes that Norwegian might be a counterexample to the
claim that the resultative adjective cannot be inflected:
() Norwegian; Åfarli (: footnote ), in Kratzer (: )
a. Vi vaska golvet rein-t/ *rein.
we washed floor.the.N.SG clean-N.SG/ clean
‘We washed the floor clean.’
b. Vi vaska rein(-t) golvet.
We washed clean floor.the.N.SG
‘We washed the floor clean.’
c. Golvet er rein-vaska/*reint-vaska.
floor.the.N.SG is clean-washed.
‘The floor is washed clean.’
In the above examples the adjective must bear inflection (see ()a) if it is not
adjacent to the verb. It optionally bears inflection when adjacent to a finite verb
(see ()b) and it cannot bear it when left-attached to the verb (see ()c). Kratzer
observes that when the adjective is overtly incorporated into the verb, as in ()b and
()c, the inflection disappears, and when it is—under her assumptions—covertly
incorporated, as in ()a, inflection is compulsory.9 This author takes the data as
suggesting that agreement morphology in (a) and (b) is a PF phenomenon, orthog-
onal to the incorporation of the adjective into CAUSE.
As was shown in section .. and already observed by Whelpton () in his
evaluation of Kratzer’s () proposal from the Icelandic perspective, Icelandic
resultatives are also built on obligatorily inflected adjectives when the adjective is
not prefixed to the verb. Furthermore, recall from section .. that resultative
adjectives in Finnish and Hungarian, although they do not bear agreement inflection,
must be endowed with a special case, which is translative in Finnish and sublative in
Hungarian. This would also be a problem for Kratzer’s proposal in the same way as is
8
Kratzer adopts Hay’s () contention that derivational affixes that can be easily parsed out should
never occur closer to the root than those that are less easily parsed out (see also Hay ). CAUSE, being
null and, hence, ranking lowest on the parsability scale, should always affix before any other (overt) affix is
added, least of all if the affix is inflectional, as agreement affixes are. But this condition can never be met
when the adjective already bears inflection before raising to CAUSE.
9
I recall that it is argued that the adjective incorporates into an upper null CAUSE head. In ()a it
remains overtly in situ, after the object; in ()b and ()c it overtly incorporates, but the linearization with
respect to the verb is different because of the presence of voice features in ()b versus their absence in ()c.
See Kratzer (: , footnote ), for a detailed explanation.
Previous approaches and possible counterexamples
Lastly, and also importantly for present concerns, Kratzer claims that adjectival
resultatives have to be studied as a phenomenon independent from similar construc-
tions involving a particle, or, as in German, a separable prefix:
10
See also Horrocks and Stavrou (: –) for a revision of Kratzer (), fundamentally on the
basis of compounding in Modern and Ancient Greek.
Previous approaches and possible counterexamples
properties restricted, in this case to PPs. This would be why Slavic does not allow AP
resultative constructions. I see two problems with this account. First, if the prefix
originates as R, it is not easy to see how it may be interpreted as a final location or a
resulting state, as argued for by Žaucer (, ), Arsenijević (), and Gehrke
(), and in this book. Rather, it seems that the prefix should be first merged as an
adposition heading the phrase which is complement to R, or as the Ground itself. The
other objection has to do with one prediction made by Svenonius’s analysis. Sveno-
nius fails to link the obligatory prefixation of R in Slavic with the fact that these
languages do not allow AP resultatives. As a result, in principle, nothing impedes the
existence of languages with an R that selects only PPs, as does, by hypothesis, Slavic,
but which is not prefixed onto the verb. Those languages would present the gram-
maticality pattern illustrated in ():
() Complex resultative predicates in a language with a non-affixal, PP-selecting R
Sue danced into the room. (Complex resultative construction based on PPs)
Sue ran in. (Complex resultative construction based on (non-affixal) particles)
*Sue beat the metal flat. (Complex resultative construction based on APs)
If that pattern is not empirically attested, which to my knowledge it is not (see section
...), Svenonius’s analysis fails to predict it.
This evidence shows that the key-AP does not qualify as a true resultative secondary
predicate sitting inside the vP, and, hence, that the constructions claimed by Son and
Svenonius () as complex AP resultatives in fact are not. This state of affairs is
compatible with Korean being a v-framed language, as already stated by Talmy
(: ).11
... Hebrew and Javanese: presence of CDMCs, absence of complex adjectival
resultatives Hebrew and Javanese are presented by Son () and Son and
Svenonius (), respectively, as languages allowing CDMCs and disallowing adjec-
tival resultatives. At first sight, this scenario is not problematic for my present
account, since Latin and Slavic have also been correctly predicted to behave in that
way. However, Hebrew and Javanese, unlike Latin and Slavic, show no signs of a
morphological dependence of the verb and the element expressing the Core Schema.
If they do allow CDMCs and their Path is not affixal, they should behave as strong
s-framed languages, like Germanic, allowing adjectival resultatives, contrary to data
presented by Son () and Son and Svenonius ().
Beginning with Hebrew, Son () reports the following scenario:
() Hebrew; Son (: )
a. *Hu kara et ha-xavila ptuxa.
he tore ACC the-package open
‘He tore the package open.’
b. *Hu cava et ha-kir adom.
he painted ACC the-wall red
‘He painted the wall red.’
() Hebrew; Son (: )
a. David {rac/zaxal} {la-xeder/ el ha-xeder}.
David ran/crawled DAT.DEF-room/ ALL the-room
‘David ran/crawled to the room.’
b. Ha-bakbuk caf {la-me’ara/ el ha-me’ara}.
the-bottle floated DAT.DEF-cave/ ALL the-cave
‘The bottle floated (in)to the cave.’
As for the predicates in (), Son () does not provide explicit aspectual tests to
show that they are telic, that is, that they qualify as true CDMCs in the sense
described in section ..., and neither do Son and Svenonius (), although
they too consider Hebrew to license CDMCs. As it turns out, Horrocks and Stavrou
(: ) note that ‘Beck and Snyder (b) show that an in-PP modifier is not
11
See Son (), and references therein, for more considerations on resultatives in Korean. I shall not
consider her work here, since she does not provide a rebuttal of Shim and Den Dikken’s () claims on
the status of the suffix -key.
A revision of Talmy’s typology
than rac ‘ran’ in ()a—the Italian counterpart of Hebrew, è gattonato, is also allowed
in directed motion constructions:
() Italian; Folli and Ramchand (: )
Il bambino di Gianni è gattonato a casa.
the child of Gianni is crawl.PTCP.PFV.M.SG at home.
‘Gianni’s child crawled home.’
However, when the construction in ()a is used with the root r-k-d ‘dance’ most of
my informants find the construction very odd or straightforwardly ungrammatical:
() Hebrew
*/??David rakad la-xeder
David danced DAT.DEF-room
‘David danced to the room.’
This is what we expect under the conjecture that Hebrew is in fact more similar to v-
framed Italian than to s-framed English. Thus, Folli and Ramchand (: ) report
that the Italian equivalent (danzare) is not possible in a goal construction. The same
obtains in other Romance languages:
() Catalan
*En Joan ha ballat a l’habitació. (Directional.)
the Joan has danced at the=room
() Spanish
*Juan ha bailado a la habitación. (Directional.)
Juan has danced at the room
As for example ()b, Asaf Bachrach, in a personal communication, informs me that
it sounds strange when accompanied by a le-PP (not by an el-PP). Moreover, the
next example in Son (: )—which I have enlarged with an in-adverbial to
ascertain its telicity—was judged by most of my informants as ungrammatical, and
Noam Faust pointed out that he needed the directional el before mitaxat ‘under’ to
render it possible:
() Hebrew; an enlarged example in Son (: )
*Ha-bakbuk caf mitaxat le-gesher tox shloshim shniyot.
the-bottle floated under DAT-bridge in thirty seconds
‘The bottle floated under the bridge in thirty seconds.’
Interestingly, another native informant reported that the only interpretation com-
patible with this example is one in which the floating of the bottle under the bridge
starts after seconds have passed, and not that the bottle floats for seconds until
it reaches the position under the bridge. Specifically, she spontaneously construed a
A revision of Talmy’s typology
compatible scene in which the bottle is held under the water and then released, taking
seconds to come to the surface and begin floating. This shows that Hebrew cannot
construe an accomplishment reading with the manner-of-motion verb equivalent to
float and a PP like mitaxat le gesher ‘under the bridge’. Rather, the frame adverbial
tox shloshim shniyot ‘in seconds’ can only have a start-time interpretation,
revealing that the predicate behaves as an activity, and not as a transition event
(see MacDonald ). This fact is in accordance with the hypothesis that Hebrew in
fact behaves like v-framed Italian, where galleggiare ‘float’ is also strange in a directed
motion construction, presumably because the root GALLEGGI is difficult to coerce into
a change-of-state reading—in my terms, it does not fit well as a Terminal Ground, in
Compl-Place:
() Italian; Folli and Ramchand (: )
*La barca è galleggiata sotto il ponte.
the boat(F) is float.PTCP.PFV.F.SG under the bridge.
‘The boat floated under the bridge.’
To sum up, if the qualifications just made on Son’s () data are on the right track,
Hebrew would behave like v-framed Romance, and not like s-framed Germanic: it
displays a wide range of path-verbs (cf. ()), it may mimic CDMCs with what
probably may correspond to a toward-like preposition (el), and, finally, it features
directed motion constructions in which the root of the verb is not really inserted as a
Co-event component adjoined to v, but, rather, as Compl-Place and must thereby be
interpreted as a Terminal Ground. This, of course, is pragmatically not possible for
every root, as exemplified with caf ‘floated’ in () and rakad ‘danced’ in (). In
conclusion, if Hebrew really turns out to be a v-framed language, its disallowance of
adjectival resultative constructions is, within the current framework, expected
(see () above).
According to Son and Svenonius (), Javanese (and Indonesian) does not allow
AP resultatives (see ()a), but does apparently allow CDMCs where the manner-of-
motion verb does not bear any affix conveying the final location of movement
(see ()b):
() Javanese; Son and Svenonius (: )
a. Mary nyacah daging *(sampek) ajur.
Mary beat meat until flat
‘Mary beat the meat until it became flat.’
b. Tika fmlaku/mlayu/mbrangkangg ning ngisor jembatan.
Tika walk/run/crawl LOC bottom bridge
‘Tika walked/ran/crawled under the bridge.’ (Both locative and directional
readings.)
Summary
The problem these data represent is the same as that discussed above with reference
to Hebrew: if Javanese is a v-framed language, ()a is expected, but not ()b. On the
other hand, if it is an s-framed language it is not clear why ()a should be out, since
there does not seem to be any morphological requirement for the result-conveying
element to be attached to the verb, as happens in Latin and Slavic. As is also the case
with Hebrew, Javanese could turn out to be a v-framed language, despite appear-
ances. In particular, two of the manner-of-motion verbs in ()b, mlayu ‘ran’ and
mbrangkangg ‘crawled’ belong to the run-class, that is, to the class of verbs that can
head change-of-state predicates in other v-framed languages like Italian or Catalan.
. Summary
In this chapter I have explored the empirical validity of the Split S-framedness
Hypothesis stated in Chapter , characterizing Ancient Greek as a weak s-framed
language and different Germanic languages and also Finnish and Hungarian as
strong s-framed. A typology has thus emerged, more fine-grained than that put
forward by Talmy, and based on the morphological properties of the Path head,
which may or may not lead to a univerbation of the transition vP and which may
interfere with independent factors of the language, such as agreement morphology
on the element expressing the result of the complex event. Already pointed out in
Chapter , in relation to Latin and Slavic, this interference has been explored further
in this chapter. Thus, in a strong s-framed language such as Icelandic, the resultative
adjective may appear prefixed to the verb only if it does not bear inflection. Com-
plementarily, weak s-framed languages like some varieties of Mandarin Chinese,
although requiring univerbation, as in Latin or Slavic, admit all kinds of complex
resultative constructions, since the element expressing result is never inflected for
agreement. Finally, I have critically examined previous accounts that focus on the
issue of the category of the resultative predicate, and I have tried to solve some of the
puzzles they involve for my own account.
7
In this chapter I concentrate on two of the principal endeavours to which I hope this
work has contributed. On the one hand, the endeavour of accurately describing the
cross-linguistic variation involved in the expression of complex transition events. On
the other hand, the endeavour to reduce cross-linguistic variation in argument
structure to how PF interprets the same syntactic output. For either case, after
pointing out the main results arrived at in the book, I discuss the challenges that
my account, together with the field in a broader perspective, faces. Thus, I examine
the properties of Complex Effected Object Constructions, whose distribution correl-
ates with Talmy’s () distinction between the class of s-framed languages and that
of v-framed languages, but which plausibly do not involve the functional head that
has been considered responsible for that typological divide, namely Path. On the
other hand, I point out a challenge in the theory of the syntax-morphology interface
adopted in this work and shared in the framework of Distributed Morphology,
namely how to handle the relationship between number of PF cycles and number
of words. In relation to this issue I examine one of the Vocabulary Items proposed for
Path in weak satellite-framed languages like Latin.
of research represented by Klipple (), Mateu (), Mateu and Rigau (),
and Real Puigdollers (, ), among others. In particular, if the realization of
Path depends on that of v, this eventive head cannot be combined with a Co-event
root. However, as pointed out in section .., and developed at length in Acedo-
Matellán and Mateu (), there is another major line of research on the type of
variation dealt with in this book, namely that initiated by Snyder (). Indeed, a
range of authors such as McIntyre (), Zubizarreta and Oh (), and Mateu
() have adopted the idea that the free use of verbs expressing manner in
constructions encoding a transition in s-framed languages is due to the availability
of the compounding of the eventive head with a manner root. Thus, Mateu (:
), for instance, argues that languages like English are characterized by allowing
‘compounding of a root with a null light verb during the syntactic derivation’. The
main reason to adopt a Snyderian rather than a Talmian perspective on the cross-
linguistic variation at hand is the existence, only in languages allowing a ‘Co-event
conflation pattern’, of constructions showing this pattern but arguably not any
directional/resultative component:
() Mateu (: )
a. John smiled his thanks.
b. The factory horns sirened midday.
In the rest of this section I will show that these constructions were also available in
Latin, and I will examine their properties. I will then show, following Mateu (,
), that they are absent in Romance, and I will discuss the theoretical implications
involved.
1
Levinson (: ) introduces the difference between explicit creation verbs and implicit creation
verbs. In the former, an example of which could be bake (a cake), the created object is expressed as an
argument of the verb, while in the latter the created object appears to be the very root of the verb. Thus, in
Mary braided her hair, a braid is entailed to be created when the event comes to conclusion, but an actual
braid is not expressed as an argument of the verb. Here I will only deal with explicit creation predicates.
Challenges and prospects
The Latin examples following are cases of CEOCs and are a further argument in
favour of aligning Latin with s-framed languages, rather than v-framed ones. The last
three of them are adapted from Lemaire ()—see also Haudry () for relevant
remarks on sterno ‘strew’:
() Latin; Cic. Fin. , ,
Qui alteri misceat mulsum.
who.NOM another.DAT mix.SBJV.SG honeyed_wine.ACC
‘He who makes honeyed wine for someone else’.
() Latin; Cic. Mil.
Vulnus [ . . . ] quod acu punctum.
wound(N)NOM.SG which.NOM.N.SG needle.ABL puncture.PTCP.PFV.NOM.N
videretur.
seem.IPFV.SBJV.SG
‘A wound that seemed to have been punctured with a needle’.
() Latin; Ov. Met. ,
[Serpens] volubilibus squamosos
snake.NOM looping.ABL.M.PL scaly.ACC.M.PL
nexibus orbes torquet.
writhing(M)ABL.PL coil(M)ACC.PL twist.SG
‘The snake twists his scaly coils in looping writhings.’
() Latin; Liv. , ,
Viam silice sternendam [ . . . ] locauerunt.
way(F)ACC flint-stone.ABL strew.PTCP.FUT.PASS.ACC.F establish.PRF.PL
‘They established that the way was to be paved with flint stone.’
() Latin; Stat. Theb. ,
Aeriam truncis [ . . . ] cumulare pyram.
high.ACC.F log.ABL.PL gather.INF pyre(F)ACC
‘To build a high pyre out of logs’.
In all these examples the verb is used as a manner modification of a creation
event. Thus, in (), the DP mulsum ‘mixed wine’ is not mixed with anything, but
is rather the result of a mixing process, and, hence, does not exist before that
process. It is crucial to bear in mind that mulsum refers to a mixture of liquids
(specifically, wine and honey), in contrast to merum, which means ‘pure,
unmixed wine’: mulsum denotes, undoubtedly, the result of the event specified
by the verb, namely, mixing. In the same way, a wound (vulnus) appears
through puncturing (see ()), the snakes’ coils (orbes) appear through twisting
(see ()), the way (viam) is created by a strewing action (see ()) and a pyre
(pyram) is created by accumulating (trunks) (see ()). Importantly, there is a
The locus of cross-linguistic variation: Talmy or Snyder?
non-creation use of all these five verbs, which does not elicit the effected object
interpretation:
() Change-of-state uses of the verbs in () to ()
a. Latin; Hor. Sat. , ,
Surrentina [ . . . ] miscet faece Falerna vina.
Surrentine.ACC.N.PL mix.SG dregs.ABL Falernian.ABL wine(N)ACC.PL
‘He mixes Surrentine wines with Falernian dregs.’
b. Latin; Cels. ,
Cutis debet [ . . . ] acu pungi.
skin.NOM must.SG needle.ABL puncture.INF.PASS
‘The skin must be punctured with a needle.’
c. Latin; Ov. Met. ,
Stamina pollice torque.
yarn.ACC thumb.ABL wind.IMP.SG
‘Spin the yarn with your thumb.’
d. Latin; Liv. , ,
Sternunt corpora.
strew.PL body.ACC.PL
‘They lay their own bodies down.’
e. Latin; Liv. , ,
Vivi mortuis [ . . . ] cumularentur.
alive.NOM.M.PL dead.DAT.M.PL heap.IPFV.SBJV.PASS.PL
‘Those alive would heap up onto the dead.’
This double use of the verbs can be argued to constitute a case of elasticity of
the verbal meaning, akin to that shown by English bake, also usable as a creation
and as a change-of-state verb (see Atkins et al. ; Pustejovsky , ; and
Mateu ).
Re-prefixation allows us to explore further the semantics of CEOCs in Latin, as will
be done for English in the next section. Consider the following example:
() Latin; Verg. Aen. ,
Re-coquont patrios fornacibus enses.
re-forge.PL paternal.ACC.M.PL furnace.ABL.PL sword(M)ACC.PL
‘They forge the forefathers’ swords anew in the furnaces.’
In this example—where COQU, referring to the submission of an object to the action of
fire, means ‘forge’—a repetitive reading of re- involving two forgings of the same
(token) swords is impossible. Specifically, we must understand that new tokens of the
same type of sword are created as a result of a forging event. The verb is, therefore,
not used as a change-of-state predicate.
Challenges and prospects
According to this parameter, the structures of () and () are impossible in languages
like Romance because they involve the combination of v with a root. However, this
account is not free of problems. From a theoretical point of view, and, particularly,
under the perspective that cross-linguistic differences stem solely from properties of the
lexicon, it is not clear why v-framed languages should disallow the combination of these
two elements. From an empirical perspective, this analysis precludes the possibility that
in v-framed languages there are instances of such combinations. Specifically, I have
Challenges and prospects
argued in section .., that such combinations are possible when there is no Path
involved, in existential predicates with a locative expression:
() Catalan; Mateu (: )
En aquesta coral n’hi canten molts, de nens.
in this choir PARTVE=LOC sing..PL many.PL of child.PL
‘There are many children who sing in this choir.’
The fact that it is in precisely these plausibly Path-less constructions that conflation is
allowed in v-framed languages suggests, on the contrary, that the v-/s-framed
distinction is linked to the presence of Path and its expression.
Future accounts of the cross-linguistic variation dealt with by Talmy and Snyder
must reconcile both views and find a unifying explanation that overcomes the
problems pointed out above.
rotolare ‘roll’ possesses a þR feature while galleggiare ‘float’ or ballare ‘dance’ does
not, which raises the question of whether this is a clear tendency in all languages and
not a quirk of the Italian lexicon.
There is another dimension in which the present approach contrasts with those
developed within the nanosyntactic framework (Starke ). As pointed out by
Borer (: ff.), in works such as the above-mentioned the very notion of
substantive terminal or root disappears. Rather, substantive lexical items are stored
chunks of structure to be inserted into a stretch of functional nodes at Spell-Out. In
my approach, however, there is a semantic motivation for locating the root contrib-
uting the relevant exponent at a given node. See the discussion in section .. and
also in Acedo-Matellán and Mateu ().
Another significant feature of the analysis of cross-linguistic variation presented in
section .. is the assumption that derivations can crash at PF. Such a crash stems
from the failure to insert the exponent of a given node, a failure caused by, first, the fact
that the node does not possess a default exponent and, second, by the fact that the
particular environment in which the node appears does not match the insertion frame
of (any of) its Vocabulary Items. The crashing character of the theory of the syntax-
morphology interface adopted in this work is not shared by other DM-based theories
but is common to Nanosyntax (cf. Fábregas’s Exhaustive Lexicalization Principle).
Maybe the main challenge to be addressed in theories assuming the realization of
individual terminals, like that adopted here, has to do with the relationship between
Spell-Out domains and words. Indeed, it does not seem the case that a given Spell-Out
domain, say that defined by the cyclic head v, always yields one single word. Embick
() does not address this problem, beyond pointing out that complex heads are
yielded either by syntactic head movement or by certain PF operations (Embick :
–). However, this cannot be the whole story, even in the cases discussed by this
author, such as the realization of inflectional material in the verb. In particular,
certain functional heads of the same v cycle may be realized as either affixes or words.
For instance, in Greek T is realized as an affix in the present tense (-o, -is, -i, etc.) but
as a non-affixal unit (a proclitic; cf. Oostendorp ) in the future tense: tha. To
compound the problem, it is not the case that v does not raise to T when T is future,
as shown by the fact that Greek tha, unlike English will/shall, does not allow the
intervention of adverbs between it and the verb:
() Greek; based on Rivero (: )
O Yánis tha (*akómi) milái (akómi).
the Yánis FUT still speak.SG still
(Cf. English Yánis will still speak.)
According to a standard analysis, the verb moves to the future T by head movement
(Alexiadou and Anagnostopoulou ). However, this movement clearly does not
feed affixation of T to v when T is future.
Challenges and prospects
The issue affects, of course, the kind of data dealt with in this book. For example,
I have claimed that in strong s-framed languages like English the same cycle headed
by v can yield either one or more than one word:
() One word vs more than one word for the v Spell-Out cycle
a. Sue [v-cycle flattened] the metal.
b. Sue [v-cycle hammered flat] the metal.
Strong s-framed languages have been characterized as not requiring the affixation of
the Path head onto v, as in weak s-framed and v-framed languages. This circum-
stance is precisely what explains the existence of complex AP resultatives such as that
in the above example. So, if Path is not specified to attach to v, the question arises
why it does so, seemingly, in ()a and not in ()b. One way to go is to stipulate that,
universally, roots raise as high as they can, that is, to the highest functional node that
is not associated itself with a root. This means that in ()a, but not in ()b, all the
nodes c-commanded by v must successively raise onto v. Place and Path receive a zero
exponent, so the exponent of v can be determined by that of the root: it is en when the
root is FLAT and it is a null exponent when the root is, for instance, MELT (yielding melt).
In ()b there is raising as far as Path, since v is already associated with a root that
raises onto it, yielding hammer. However, this explanation does not straightforwardly
account for the cases in which v is realized as a (non-affixal) light verb:
() Sue got the metal flat.
The stipulation introduced above forces the root FLAT to end up affixed onto v, which
precludes the insertion of the exponent get. If we maintain that there are, in fact, light
verbs, and that get does not involve a root, the above stipulation could be eliminated
in favour of a laissez-faire scenario in which the root can either raise as high as v or
not. If it does not raise, get is inserted into v, as a last resort option. If the root raises, it
yields the synthetic verb. In future research on the issue of the synthetic/analytic
expression of PF cycles it will be crucial, therefore, to explore the extent to which
roots universally raise to the c-commanding functional heads.2
The issue of the raising of roots to the upper functional heads is also involved in a
plausible, more natural derivation of Latin complex predicates headed by prefixed
verbs and of the ban on unprefixed complex resultative constructions in this lan-
guage. In section .. I argued for the next Vocabulary Item for Path in languages
like Latin and Slavic (weak satellite-framed languages):
() Path $ ∅ / Place-_-[ . . . v . . . ]
2
See Acedo-Matellán () for an analysis of French expressions like Avoir froid/faim ‘Have cold/
hunger’ in which the apparent noun (froid ‘cold’, faim ‘hunger’) is shown to be a non-incorporated root.
Reducing cross-linguistic variation to PF properties
This Vocabulary Item forces the raising of all the material in the PathP to v,
accounting for the shape of complex resultative constructions in these languages
(prefixed) and for the non-existence of complex adjectival or PP resultatives
based on unprefixed verbs. Thus, unprefixed resultative constructions are ruled
out in the language because they involve a PlaceP that, by virtue of agreement
morphology, is spelled out as an independent cycle. The insertion frame of the
Vocabulary Item for Path, which requires that Place linearly precede this head, is
not met and the derivation crashes. On the other hand, I have shown that at
least in Latin there is a case in which Path is prefixed onto the verb without the
rest of the material in PathP (Place and the root), namely, when this head is
strictly left-adjacent to v, in simple resultative constructions (see section ..).
In this case, Path receives an overt exponent (re) according to the following
Vocabulary Item:
() Path $ re /_ -v-Voice
However, there may be a more principled way to derive the same empirical results
while reducing the difference between these Vocabulary Items. Specifically,
I speculate that the Vocabulary Item in () can be simplified by eliminating from
the insertion frame the requirement that Place linearly precede Path:
() Path $ ∅ / _-[ . . . v . . . ]
If () and () are the two Vocabulary Items for Path in Latin, it can be stated that
Path must linearly precede v in this language, the difference being between an
immediate (re) or a not immediate (a null exponent) linear precedence. However,
it is true that, with no other proviso, () allows the generation of complex adjectival
resultative constructions in Latin, contrary to fact (see section ..):
() Latin
*Ovidia poculum vacu-um bibit.
Ovidia.NOM goblet(N)ACC.SG empty-ACC.N.SG drink.SG
b. Raising
[v Path [v BIB v]]
c. Linearization
Path-BIB-v
d. Vocabulary Insertion
∅-bib-∅
Challenges and prospects
() Liv. , ,
Quod ubi videre ipsum Camillum, [ . . . ] vadentem in hostes, pro-currunt pariter omnes.
‘As soon as they see Camillus marching against the enemy, all of them run forth in like
fashion.’
() Curt. , ,
Quod ubi exercitus [ . . . ] conspexit, [ . . . ] con-currit.
‘The army ran up to the place as soon as they spotted this.’
() Curt. , ,
Subito [ . . . ] rex Indus [ . . . ] oc-currit.
‘Suddenly the king Indus runs to meet them.’
() Cels. ,
Ne suc-currere quidem statim sibi possunt.
‘They are not able to assist themselves immediately.’
() Cels. ,
Subito nigra alvus pro-fluxit.
‘Suddenly a black flux flows forth.’
() Colum. ,
Deinde ubi liquatum mel in subiectum alveum de-fluxit, transferetur in vasa fictilia.
‘Then as soon as the liquefied honey has flowed completely into the vessel located
underneath, it should be transferred into earthenware vessels.’
() Vitr. , ,
Tiburtina [ . . . ] simul [ . . . ] sunt ab [igne] [ . . . ] tacta, dis-siliunt et dissipantur.
‘The Tiburtine stones, as soon as they have been touched by fire burst asunder and
scatter.’
() Curt. , ,
Ut primum rex in conspectu fuit, equo ipsa de-siluit.
‘As soon as she saw the king she leapt down from the horse herself.’
() Curt. , ,
Singuli repente de-siliunt.
‘They suddenly leap down one by one.’
() Curt. , ,
Ex lecto repente pro-siluit.
‘He suddenly leapt forth from the bed.’
() Phaedr. , ,
Subito latrones ex insidiis ad-volant.
‘Suddenly the thieves fly onto them out of the ambush.’
Alexiadou, Artemis and Elena Anagnostopoulou (). Parametrizing AGR: Word order,
V-movement and EPP-checking, Natural Language & Linguistic Theory : –.
Allen, Margaret (). Morphological Investigations. PhD thesis, University of Connecticut.
Ambridge, Ben and Silke Brandt (). Lisa filled water into the cup: The roles of entrench-
ment, pre-emption and verb semantics in German speakers’ L acquisition of English
locatives, Zeitschrift für Anglistik und Amerikanistik : –.
Amritavalli, Raghavachari (). Syntactic categories and lexical argument structure, in Eric
Reuland, Tanmoy Bhattacharya, and Giorgos Spathas (eds.), Argument Structure. Amster-
dam: John Benjamins, –.
Amritavalli, Raghavachari and Karattuparambil A. Jayaseelan (). The genesis of syntactic
categories and parametric variation, in Huang-Jin Yoon (ed.), Proceedings of the th GLOW
in Asia . Generative Grammar in a Broader Perspective. Seoul: The Korean Generative
Grammar Circle and Institute for Cognitive Science, Seoul National University, –.
An, Duk-Ho (). Clauses in noncanonical positions at the syntax-phonology interface,
Syntax : –.
Anderson, Stephen (). On the role of deep structure in semantic representation, Founda-
tions of Language : –.
Anderson, Stephen (). Comments on the paper by Wasow, in Peter Culicover, Thomas
Wasow, and Adrian Akmajian (eds.), Formal Syntax. New York: Academic Press, –.
Arad, Maya (). A minimalist view of the syntax-lexical semantics interface, UCL Working
Papers in Linguistics : –.
Arad, Maya (). VP-Structure and the Syntax-Lexicon Interface. PhD thesis, University
College London.
Arad, Maya (). Locality constraints on the interpretation of roots: The case of Hebrew
denominal verbs, Natural Language & Linguistic Theory : –.
Arad, Maya (). Roots and Patterns. Hebrew Morpho-syntax. Dordrecht: Springer.
Arregi, Karlos and Andrew Nevins (). Morphotactics: Basque Auxiliaries and the Structure
of Spellout. Dordrecht: Springer.
Arsenijević, Boban (). Inner Aspect and Telicity: The Decompositional and the Quantifica-
tional Nature of Eventualities at the Syntax-Semantics Interface. Utrecht: LOT Publications.
Aske, Jon (). Path predicates in English and Spanish: A closer look, Berkeley Linguistics
Society : –.
Atkins, Beryl T., Judy Kegl, and Beth Levin (). Anatomy of a Verb Entry: From Linguistic
Theory to Lexicographic Practice, International Journal of Lexicography : –.
Babko-Malaya, Olga (). Zero Morphology: A Study of Aspect, Argument Structure Case.
PhD thesis, State University of New Jersey.
Baker, Mark C. (). Incorporation: A Theory of Grammatical Function Changing. Chicago:
University of Chicago Press.
Baker, Mark C. (). Thematic roles and syntactic structure, in Liliane Haegeman (ed.),
Elements of Grammar. Dordrecht: Kluwer, –.
Baker, Mark C. (). Lexical Categories. Verbs, Nouns and Adjectives. Cambridge: Cam-
bridge University Press.
Barbelenet, Daniel (). De l’aspect verbal en latin ancien et particulièrement dans Terence.
Paris: Honoré Champion.
References
Bauke, Leah S. (). Symmetry Breaking in Syntax and the Lexicon. Amsterdam: John
Benjamins.
Beavers, John (). Argument/Oblique Alternations and the Structure of Lexical Meaning.
PhD thesis, Stanford University.
Beavers, John, Beth Levin, and Shiao Wei Tham (). The typology of motion expressions
revisited, Journal of Linguistics : –.
Beck, Sigrid and William Snyder (a). Complex predicates and goal PP’s: Evidence for a
semantic parameter, Boston University Conference on Language Development : –.
Beck, Sigrid and William Snyder (b). The resultative parameter and restitutive again, in
Caroline Féry and Wolfgang Sternefeld (eds.), Auditur vox sapientiae: a Festschrift for Arnim
von Stechow. Berlin: Akademie Verlag, –.
Bende-Farkas, Ágnes (). Complex predicates and discourse referents: The case of Hun-
garian. Handout of talk given at the Workshop on Predicative Constructions, Berlin,
October.
Bennis, Hans (). Theoretische aspecten van partikelvooropplaatsing, Tabu : –.
Berman, Ruth A. and Yonni Neeman (). Development of linguistic forms: Hebrew, in
Ruth A. Berman and Dan I. Slobin (eds.), Relating Events in Narrative. A Crosslinguistic
Developmental Study. Hillsdale, NJ: Lawrence Erlbaum Associates, –.
Binnick, Robert (). Studies in the Derivation of Predicative Structures. PhD thesis, Uni-
versity of Chicago.
Biskup, Petr and Michael Putnam (). One P with two Spell-Outs: the ent/aus-alternation
in German, Linguistic Analysis : –.
Boas, Hans C. (). Resultative Constructions in English and German. PhD thesis, University
of North Carolina.
Boas, Hans C. (). A Constructional Approach to Resultatives. Stanford, CA: CSLI.
Boeckx, Cedric (). Elementary Syntactic Structures: Prospects of a Feature-Free Syntax.
Cambridge: Cambridge University Press.
Bohnemeyer, Jürgen and Mary Swift (). Event realization and default aspect, Linguistics
and Philosophy : –.
Bolkestein, Machtelt (). The ‘Ab urbe condita’-construction in Latin: A strange type of
raising?, in Saskia Daalder and Marinel Gerritsen (eds.), Linguistics in the Netherlands .
Amsterdam: North-Holland, –.
Bolkestein, Machtelt (). Parameters in the expression of embedded predications in Latin,
in Gualtiero Calboli (ed.), Subordination and other Topics in Latin. Proceedings of the Third
Colloquium on Latin Linguistics, Bologna, – April . Amsterdam: John Benjamins,
–.
Bonet, Eulàlia (). Morphology after Syntax: Pronominal Clitics in Romance. PhD thesis,
Massachusetts Institute of Technology.
Booij, Geert (). The boundary between morphology and syntax: separable complex verbs
in Dutch, Yearbook of Morphology : –.
Booij, Geert (). Periphrastic word formation, in Geert Booij, Janet DeCesaris, Angela Ralli,
and Sergio Scalise (eds.), Topics in Morphology: Selected Papers from the Third Mediterra-
nean Morphology Meeting, Barcelona, September –, . Barcelona: Institut Universi-
tari de Lingüística Aplicada (Universitat Pompeu Fabra), –.
References
Borer, Hagit (). Deriving passives without theta-grids, in Steven G. Lapointe, Diane
K. Brentari, and Patrick M. Farrell (eds.), Morphology and its Relations to Phonology and
Syntax. Stanford, CA: CSLI, –.
Borer, Hagit (). Exo-skeletal vs. endo-skeletal explanations: Syntactic projections and the
lexicon, in John C. Moore and Maria Polinsky (eds.), The Nature of Explanation in Linguistic
Theory. Stanford, CA: CSLI, –.
Borer, Hagit (a). Structuring Sense, vol. : In Name Only. Oxford: Oxford University Press.
Borer, Hagit (b). Structuring Sense, vol. : The Normal Course of Events. Oxford: Oxford
University Press.
Borer, Hagit (). Structuring Sense, vol. : Taking Form. Oxford: Oxford University Press.
Borer, Hagit (). Wherefore roots?, Theoretical Linguistics : –.
Borik, Olga (). Aspect and Reference Time. Oxford: Oxford University Press.
Bortolussi, Bernard (). L’objet interne en latin, in Heinrich Weber and Ryszard Zuber
(eds.), Linguistik Parisette: Akten des . Linguistischen Kolloquiums, Paris . Tübingen:
Niemeyer, –.
Bortolussi, Bernard (). Le double accusatif avec les verbes préfixés en circum- et trans-, in
Claude Moussy (ed.), La composition et la préverbation en latin. Paris: Presses de l’Université
Paris-Sorbonne, –.
Bouchard, Denis (). The Semantics of Syntax. Chicago: University of Chicago Press.
Brachet, Jean-Paul (). Recherches sur les préverbes de- et ex- du latin. Brussels: Latomus.
Broekhuis, Hans (). Adpositions and Adpositional Phrases. Tilburg: Tilburg University.
Burzio, Luigi (). Italian Syntax: A Government and Binding Approach. Dordrecht: Reidel.
Bye, Patrik and Peter Svenonius (). Non-concatenative morphology as epiphenomenon,
in Jochen Trommer (ed.), The Morphology and Phonology of Exponence. Oxford: Oxford
University Press, –.
Caha, Pavel (). The German locative-directional alternation, The Journal of Comparative
Germanic Linguistics : –.
Cappelle, Bert (). Particle Patterns in English: A Comprehensive Coverage. PhD thesis,
Katholieke Universiteit Leuven.
Carlson, Gregory (). Reference to Kinds in English. PhD thesis, University of California.
Carrier, Jill and Janet H. Randall (). The argument structure and syntactic structure of
resultatives, Linguistic Inquiry : –.
Carvalho, Paulo de (). La grammaire de la préverbation, ou: préverbation et ergativité, in
Benjamín García Hernández (ed.), Estudios de lingüística latina. Actas del IX Coloquio
Internacional de Lingüística Latina. Universidad Autónoma de Madrid, – de Abril de
. Madrid: Ediciones Clásicas, –.
Centineo, Giulia (). A lexical theory of auxiliary selection in Italian, Probus : –.
Chen, Liang and Jiansheng Guo (). Motion events in Chinese novels: Evidence for an
equipollently-framed language, Journal of Pragmatics : –.
Chomsky, Noam (). Lectures on Government and Binding: The Pisa Lectures. Dordrecht:
Foris Publications.
Chomsky, Noam (). A Minimalist Program for Linguistic Theory, in Kenneth Hale and
Samuel J. Keyser (eds.), The View from Building : Essays in Linguistics in Honor of Sylvain
Bromberger. Cambridge, MA: MIT Press, –.
References
Chomsky, Noam (). The Minimalist Program. Cambridge, MA: MIT Press.
Chomsky, Noam (). Minimalist inquiries: The framework, in Roger Martin, David
Michaels, and Juan Uriagereka (eds.), Step by Step: Essays on Minimalist Syntax in Honor
of Howard Lasnik. Cambridge, MA: MIT Press, –.
Chomsky, Noam (). Derivation by phase, in Michael Kenstowicz (ed.), Ken Hale: A Life in
Language. Cambridge, MA: MIT Press, –.
Chomsky, Noam (). On phases, in Robert Freidin, Carlos P. Otero, and María Luisa
Zubizarreta (eds.), Foundational Issues in Linguistic Theory. Essays in Honor of Jean-Roger
Vergnaud. Cambridge, MA: MIT Press, –.
Citko, Barbara (). Phase Theory. Cambridge: Cambridge University Press.
Coleman, Robert (). The rise and fall of absolute constructions in Latin case history, in
Gualtiero Calboli (ed.), Subordination and other Topics in Latin. Proceedings of the Third
Colloquium on Latin Linguistics, Bologna, – April . Amsterdam: John Benjamins,
–.
Coleman, Robert (). Latin prepositional syntax in Indo-European perspective, in Robert
Coleman (ed.), New Studies in Latin Linguistics. Selected papers from the th International
Colloquium on Latin Linguistics. Cambridge, April . Amsterdam: John Benjamins,
–.
Comrie, Bernard (). Aspect: An Introduction to the Study of Verbal Aspect and Related
Problems. Cambridge: Cambridge University Press.
Crane, Gregory R. (ed.) (). Perseus Digital Library Project. Medford, MA: Tufts University.
Crocco Galèas, Grazia and Claudio Iacobini (). Lo sviluppo del tipo verbale parasintetico
in latino: prefissi AD-, IN-, EX-, Quaderni patavini di linguistica : –.
Csirmaz, Aniko (). Particles and a two component theory of aspect, in Katalin É. Kiss
(ed.), Event Structure and the Left Periphery. Studies on Hungarian. Dordrecht: Springer,
–.
Cuervo, María Cristina (). Datives at Large. PhD thesis, Massachusetts Institute of
Technology.
Damonte, Federico (). Classifier incorporation and the Locative Alternation, in Nicola
Munaro, Giuseppina Turano, and Walter Schweikert (eds.), Proceedings of the XXX Incontro
di Grammatica Generativa. Venice: Cafoscarina, –.
De Belder, Marijke (). Roots and Affixes. Eliminating Lexical Categories from Syntax.
Utrecht: LOT Publications.
De Belder, Marijke and Jeroen van Craenenbroeck (). How to merge a root, Linguistic
Inquiry (): –.
Demirdache, Hamida and Myriam Uribe-Etxebarria (). The syntax of time adverbs, in
Jacqueline Guéron and Jacqueline Lecarme (eds.), The Syntax of Time. Cambridge, MA:
MIT Press, –.
Den Dikken, Marcel (). Particles. On the Syntax of Verb-Particle, Triadic and Causative
Constructions. Oxford: Oxford University Press.
Den Dikken, Marcel (). On the syntax of locative and directional adpositional phrases.
MS, City University of New York.
Den Dikken, Marcel (). Directions from the GET-GO. On the syntax of manner-of-
motion verbs in directional constructions, Catalan Journal of Linguistics : –.
References
Gehrke, Berit (). Ps in Motion: On the Semantics and Syntax of P Elements and Motion
Events. Utrecht: LOT Publications.
Ghomeshi, Jila and Diane Massam (). Lexical/syntactic relations without projections,
Linguistic Analysis : –.
Gianollo, Chiara (). Il medio in latino e il fenomeno dell’intransitività scissa. Master’s
thesis, Università di Pisa.
Gibert Sotelo, Elisabeth (forthcoming). Source and Negative Prefixes. On the Syntax-lexicon
Interface and the Encoding of Spatial Relations. PhD thesis, Universitat de Girona.
Gleitman, Lila and Barbara Landau (). Every child an isolate: nature’s experiments in
language learning, in Massimo Piattelli-Palmarini and Robert C. Berwick (eds.), Rich
Languages from Poor Inputs. Oxford: Oxford University Press, –.
Goldberg, Adele (). Constructions. A Construction Grammar Approach to Argument
Structure. Chicago: University of Chicago Press.
Goldberg, Adele (). Constructions at Work. The Nature of Generalization in Language.
Oxford: Oxford University Press.
Green, Georgia (). Some observations on the syntax and semantics of instrumental verbs,
Chicago Linguistic Society : –.
Grimshaw, Jane (). Argument Structure. Cambridge, MA: MIT Press.
Haider, Hubert and Martin Prinzhorn (eds.) (). Verb Second Phenomena in Germanic
Languages. Dordrecht: Foris.
Hale, Kenneth (). Notes on world view and semantic categories: Some Warlpiri examples,
in Peter Muysken and Henk van Riemsdijk (eds.), Features and Projections. Dordrecht:
Foris, –.
Hale, Kenneth and Samuel J. Keyser (). The syntactic character of thematic structure, in
Iggy M. Roca (ed.), Thematic Structure: Its Role in Grammar. Berlin: Foris, –.
Hale, Kenneth and Samuel J. Keyser (). On argument structure and the lexical expression
of syntactic relations, in Kenneth Hale and Samuel J. Keyser (eds.), The View from Building
: Essays in Linguistics in Honor of Sylvain Bromberger. Cambridge, MA: MIT Press,
–.
Hale, Kenneth and Samuel J. Keyser (). Adjectives, other stative predicates and the roots of
stativity. MS, Massachusetts Institute of Technology.
Hale, Kenneth and Samuel J. Keyser (). The basic elements of argument structure, MIT
Working Papers in Linguistics : –.
Hale, Kenneth and Samuel J. Keyser (a). Bound features, Merge, and transitivity alterna-
tions, in Liina Pylkkänen, Angeliek van Hout, and Heidi Harley (eds.), Papers from the
UPenn/MIT Roundtable on the Lexicon. Cambridge, MA: MIT Working Papers in Linguis-
tics, –.
Hale, Kenneth and Samuel J. Keyser (b). A response to Fodor and Lepore, ‘Impossible
words?’, Linguistic Inquiry : –.
Hale, Kenneth and Samuel J. Keyser (). Prolegomenon to a Theory of Argument Structure.
Cambridge, MA: MIT Press.
Halle, Morris (). An approach to morphology. North-Eastern Linguistic Society :
–.
Halle, Morris (). Latvian Declension, Morphology Yearbook : –.
References
Halle, Morris (). Distributed morphology: Impoverishment and fission, MIT Working
Papers in Linguistics : –.
Halle, Morris and Alec Marantz (). Distributed Morphology and the pieces of inflection,
in Kenneth Hale and Samuel J. Keyser (eds.), The View from Building : Essays in
Linguistics in Honor of Sylvain Bromberger. Cambridge, MA: MIT Press, –.
Halle, Morris and Alec Marantz (). Some key features of Distributed Morphology, MIT
Working Papers in Linguistics : –.
Halliday, Michael A. K. (). Notes on transitivity and theme in English. Part I, Journal of
Linguistics : –.
Harley, Heidi (). Subjects, Events, and Licensing. PhD thesis, Massachusetts Institute of
Technology.
Harley, Heidi (). How do verbs get their names? Denominal verbs, Manner incorporation
and the ontology of verb roots in English, in Nomi Erteschik-Shir and Tova Rapoport (eds.),
The Syntax of Aspect. Deriving Thematic and Aspectual Interpretation. Oxford: Oxford
University Press, –.
Harley, Heidi (). The bipartite structure of verbs cross-linguistically (or why Mary can’t
exhibit John her paintings). MS, University of Arizona.
Harley, Heidi (). External arguments and the Mirror Principle: On the distinctness of
Voice and v, Lingua: : –.
Harley, Heidi (). On the identity of roots, Theoretical Linguistics : –.
Harley, Heidi and Rolf Noyer (). Licensing in the non-lexicalist lexicon: nominalizations,
Vocabulary Items and the Encyclopedia, MIT Working Papers in Linguistics : –.
Harley, Heidi and Rolf Noyer (). Distributed Morphology, Glot International : –.
Harley, Heidi and Rolf Noyer (). Licensing in the non-lexicalist lexicon, in Bert Peeters
(ed.), The Lexicon/Encyclopaedia Interface. Amsterdam: Elsevier, –.
Haudry, Jean (). L’instrumental et la structure de la phrase simple en indo-européen,
Bulletin de la Société de Linguistique : –.
Haugen, Jason D. (). Hyponymous objects and Late Insertion, Lingua : –.
Haverling, Gerd (). On the sco-suffix, on prefixes, and on the development of the Latin
verbal system, in József Herman (ed.), Linguistic Studies on Latin. Selected Papers from the
th International Colloquium on Latin Linguistics (Budapest, – March ). Amster-
dam: John Benjamins, –.
Haverling, Gerd (a). On sco-verbs, on prefixes, and on semantic functions, in Hannah
Rosén (ed.), Aspects of Latin: Papers from the Seventh International Colloquium on Latin
Linguistics, Jerusalem, April . Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft,
–.
Haverling, Gerd (b). Some more remarks on sco-verbs, on prefixes, and on semantic
functions, in Alfred Bammesberger and Friedrich Heberlein (eds.), Akten des VIII. Internatio-
nalen Kolloquiums zur Lateinischen Linguistik. Proceedings of the Eighth International Collo-
quium on Latin Linguistics. Heidelberg: Universitätsverlag C. Winter, –.
Haverling, Gerd (). On the development of the perfect and the imperfect tenses in Late
Latin, in Benjamín García Hernández (ed.), Estudios de Lingüística Latina. Actas del IX
Coloquio Internacional de Lingüística Latina. Universidad Autónoma de Madrid, – de
Abril de . Madrid: Ediciones Clásicas, –.
References
Haverling, Gerd (). On Sco-verbs, Prefixes and Semantic Functions. A Study in the
Development of Prefixed and Unprefixed Verbs from Early to Late Latin. Gothenburg: Acta
Universitatis Gothoburgensis.
Haverling, Gerd (). On prefixes and actionality in Classical and Late Latin, Acta Linguis-
tica Hungarica : –.
Hay, Jennifer (). Causes and Consequences of Word Structure. PhD thesis, Northwestern
University.
Hay, Jennifer (). Causes and Consequences of Word Structure. New York: Routledge.
Hegedűs, Veronika (). Hungarian spatial PPs, Nordlyd : –.
Heslin, Thomas P. Jr. (). Prefixation in Latin, Lingua : –.
Hewson, John (). The Latin verbal system, in John Hewson and Vit Bubenik (eds.), Tense
and Aspect in Indo-European Languages. Theory, Typology, Diachrony. Amsterdam: John
Benjamins, –.
Hoekstra, Teun (). Small clause results, Lingua : –.
Hoekstra, Teun (). Auxiliary selection in Dutch, Natural Language & Linguistic Theory :
–.
Hoekstra, Teun and René Mulder (). Unergatives as copular verbs: Locational and
existential predication, The Linguistic Review : –.
Hofmann, Johann B. and Anton Szantyr (). Lateinische Grammatik, vol. : Syntax und
Stilistik. München: C. H. Beck.
Horrocks, Geoffrey (). Aspect and verbs of movement in the history of Greek: Why
Pericles could ‘walk into town’ but Karamanlis could not, in J. H. W Penney, Indo-European
Perspectives. Studies in Honour of Anna Morpurgo Davies. Oxford: Oxford University Press,
–.
Horrocks, Geoffrey and Melita Stavrou (). Actions and their results in Greek and English:
The complementarity of morphologically encoded (viewpoint) aspect and syntactic resulta-
tive predication, Journal of Semantics : –.
Horrocks, Geoffrey and Melita Stavrou (). Grammaticalized aspect and spatio-temporal
culmination, Lingua : –.
Horvath, Julia (). Verbal prefixes: A non-category in Hungarian, Glossa : –.
Hout, Angeliek van (). Event Semantic of Verb Frame Alternations: A Case Study of Dutch
and its Acquisition. PhD thesis, Tilburg University.
Hout, Angeliek van (). Projection based on event structure, in Peter Coopmans, Martin
Everaert, and Jane Grimshaw (eds.), Lexical Specification and Insertion. Amsterdam: John
Benjamins, –.
Inagaki, Shunji (). Motion verbs with locational/directional PPs in English and Japanese,
Canadian Journal of Linguistics : –.
Irube, Kaneharu (). Argument structure and the out-prefixation, English Linguistics :
–.
Jackendoff, Ray (). X-Bar Syntax: A Study of Phrase Structure. Cambridge, MA: MIT
Press.
Jackendoff, Ray (). Semantics and Cognition. Cambridge, MA: MIT Press.
Jensen, Povl J. (). The use of the perfect in Classical Latin, Classica et Mediaevalia :
–.
References
Julien, Marit (). Syntactic Heads and Word Formation. Oxford: Oxford University Press.
Kaplan, Ronald M. and Joan Bresnan (). Lexical-Functional Grammar: A formal system
for grammatical representation, in Joan Bresnan (ed.), The Mental Representation of Gram-
matical Relations. Cambridge, MA: MIT Press, –.
Kayne, Richard (). French Syntax. Cambridge, MA: MIT Press.
Kayne, Richard (). Principles of particle constructions, in Jacqueline Guéron, Hans-Georg
Obenauer, and Jean-Yves Pollock (eds.), Grammatical Representations. Dordrecht: Foris,
–.
Kayne, Richard (). Antisymmetry and the Lexicon, Linguistic Variation Yearbook : –.
Keyser, Samuel J. and James Roeper (). Re: The Abstract Clitic Hypothesis, Linguistic
Inquiry : –.
Kiparsky, Paul (). Lexical morphology and phonology, in Linguistic Society of Korea (ed.),
Linguistics in the Morning Calm: Selected Essays from SICOL-. Seoul: Hanshin, –.
Kiparsky, Paul (). Remarks on denominal verbs, in Àlex Alsina, Joan Bresnan, and Peter
Sells (eds.), Complex Predicates. Stanford, CA: CSLI, –.
Klein, Wolfgang (). A time-relational analysis of Russian aspect, Language : –.
Klipple, Elisabeth (). Prepositions and variation, in Anna-Maria Di Sciullo (ed.), Projec-
tions and Interface Conditions: Essays on Modularity. Oxford: Oxford University Press,
–.
Kolehmainen, Leena (). Präfix- und Partikelverben im deutsch-finnischen Kontrast.
Helsinki: Helsingin yliopiston verkkojulkaisut.
Koopman, Hilda (). The Syntax of Specifiers and Heads. London: Routledge.
Kopecka, Anetta (). Étude typologique de l’expression de l’espace: localisation et déplace-
ment en français et en polonais. PhD thesis, Université Lumière Lyon .
Kratzer, Angelika (). Stage-Level and Individual-Level Predicates, in Gregory N. Carlson
and Francis Jeffry Pelletier (eds.), The Generic Book. Chicago: University of Chicago Press,
–.
Kratzer, Angelika (). Severing the external argument from its verb, in Johan Rooryck and
Laurie Zaring (eds.), Phrase Structure and the Lexicon. Dordrecht: Kluwer, –.
Kratzer, Angelika (). Building statives, Berkeley Linguistic Society : –.
Kratzer, Angelika (). Building resultatives, in Claudia Maienborn and Angelika Wöllstein-
Leisten (eds.), Event Arguments: Foundations and Applications. Tübingen: Niemeyer, –.
Krifka, Manfred (). Thematic relations as links between nominal reference and temporal
constitution, in Ivan A. Sag and Anna Szabolcsi (eds.), Lexical Matters. Stanford, CA:
Stanford University, –.
Kühner, Raphael and Carl Stegmann (). Ausführliche Grammatik der lateinischen Sprache,
vol. I: Elementar-, Formen- und Wortlehre. Hanover: Hahnsche Buchhandlung.
Kuno, Susumu and Ken-ichi Takami (). Functional Constraints in Grammar: On the
Unergative-Unaccusative Distinction. Amsterdam: John Benjamins.
Landgraf, Gustav von (). Der Dativus commodi und der Dativus finalis mit ihren Abarten,
Archiv für Lateinische Lexikographie : –.
Larson, Richard K. (). On the double object construction, Linguistic Inquiry : –.
Larson, Richard K. (). Double objects revisited: A reply to Jackendoff, Linguistic Inquiry
: –.
References
Lavency, Marius (). À propos de la question des cas en latin classique, Études Classiques
: –.
Lavency, Marius (). Le Paradigme syntaxique de l’ablatif absolu, in Freddy Decreus and
Carl Deroux (eds.), Hommages à Jozef Veremans. Brussels: Latomus, –.
Lavidas, Nikolaos (). Transitivity Alternations in Diachrony. Changes in Argument Struc-
ture and Voice Morphology. Newcastle upon Tyne: Cambridge Scholars Publishing.
Legendre, Géraldine (). Unaccusativity in French, Lingua : –.
Lehmann, Christian (). Latin preverbs and cases, in Harm Pinkster (ed.), Latin Linguistics
and Linguistic Theory. Proceedings of the st International Colloquium on Latin Linguistics.
Amsterdam, April . Amsterdam: John Benjamins, –.
Lehrer, Adrienne (). Prefixes in English word formation, Folia Linguistica : –.
Lemaire, Maryvonne (). Un aspect de l’échange des arguments du verbe: les verbes à
double régime de l’objet en latin, Bulletin de la Société de Linguistique de Paris : –.
Le Roux, Cecile (). On the Interface of Morphology and Syntax. Stellenbosch: University of
Stellenbosch.
Levin, Beth (). Verb Classes and Alternations: A Preliminary Investigation. Chicago:
University of Chicago Press.
Levin, Beth (). Objecthood: An event structure perspective, Chicago Linguistic Society :
–.
Levin, Beth and Tova R. Rapoport (). Lexical Subordination, Chicago Linguistic Society :
–.
Levin, Beth and Malka Rappaport (). Non-event -er nominals: A probe into argument
structure, Linguistics : –.
Levin, Beth and Malka Rappaport Hovav (). Unaccusativity: At the Syntax-Lexical Seman-
tics Interface. Cambridge, MA: MIT Press.
Levin, Beth and Malka Rappaport Hovav (). The semantic determinants of argument
expressions: A view from the English resultative construction, in Jacqueline Guéron and
Jacqueline Lecarme (eds.), The Syntax of Time. Cambridge, MA: MIT Press, –.
Levin, Beth and Malka Rappaport Hovav (). Argument Realization. Cambridge: Cam-
bridge University Press.
Levin, Beth and Peter Sells (). Unpredicated particles, in Linda Uyechi and Lian Hee Wee
(eds.), Reality Exploration and Discovery: Pattern Interaction in Language & Life. Stanford,
CA: CSLI, –.
Levinson, Lisa (). The Roots of Verbs. PhD thesis, New York University.
Levinson, Lisa (). Arguments for pseudo-resultative predicates, Natural Language &
Linguistic Theory : –.
Levinson, Lisa (). The ontology of roots and verbs, in Artemis Alexiadou, Hagit Borer, and
Florian Schäfer (eds.), The Syntax of Roots and the Roots of Syntax. Oxford: Oxford
University Press, –.
Lewandowski, Wojciech (). The locative alternation in verb-framed vs. satellite-framed
languages: A corpus study of Spanish and Polish, Studies in Language : –.
Lewis, Charlton T. and Charles Short (). A Latin Dictionary: Founded on Andrews’ Edition
of Freund’s Latin Dictionary: Revised, Enlarged, and in Great Part Rewritten by Charlton
T. Lewis, Ph.D. and Charles Short, LL. D. Oxford: Clarendon Press.
References
Liddell, Henry G. and Robert Scott (). A Greek-English Lexicon. Revised and Augmented
throughout by Sir Henry Stuart Jones with the Assistance of Roderick McKenzie. Oxford:
Clarendon Press.
Lieber, Rochelle (). Morphology and Lexical Semantics. Cambridge: Cambridge University
Press.
Lindvall, Ann (). Swedish particle verbs in comparison with Polish aspect marking,
Travaux de l'institut de linguistique de Lund : –.
Löfstedt, Bengt (). Zum lateinischen possessiven Dativ. Zeitschrift für vergleichende
Sprachforschung auf dem Gebiete der Indogermanischen Sprachen : –.
Lohndal, Terje (). Phrase Structure and Argument Structure. A Case Study of the Syntax-
Semantic Interface. Oxford: Oxford University Press.
López Moreda, Santiago (). Interferencias semántico-sintácticas entre preverbios y pre-
posiciones, in Benjamín García Hernández (ed.), Estudios de lingüística latina. Actas del IX
Coloquio Internacional de Lingüística Latina. Universidad Autónoma de Madrid, – de
Abril de . Madrid: Ediciones Clásicas, –.
Lüdeling, Anke (). On Particle Verbs and Similar Constructions in German. Stanford, CA: CSLI.
Luraghi, Silvia (). The relationship between prepositions and cases within Latin prepos-
itional phrases, in Gualtiero Calboli (ed.), Subordination and other Topics in Latin. Proceed-
ings of the Third Colloquium on Latin Linguistics, Bologna, – April . Amsterdam: John
Benjamins, –.
Luraghi, Silvia (). Omission of the direct object in Classical Latin, Indogermanische
Forschungen : –.
MacDonald, Jonathan E. (). The Syntactic Nature of Inner Aspect. A Minimalist Perspec-
tive. Amsterdam: John Benjamins.
McIntyre, Andrew (). German Double Particles as Preverbs: Morphology and Conceptual
Semantics. Berlin: Stauffenberg.
McIntyre, Andrew (). Idiosyncrasy in particle verbs, in Nicole Dehé, Ray Jackendoff,
Andrew McIntyre, and Silke Urban (eds.), Verb-Particle Explorations. Berlin: Mouton de
Gruyter, –.
McIntyre, Andrew (). Preverbs, argument linking and verb semantics: Germanic prefixes
and particles, Yearbook of Morphology : –.
McIntyre, Andrew (). Event paths, conflation, argument structure, and VP shells, Lin-
guistics : –.
McIntyre, Andrew (). Particle verbs and argument structure, Language and Linguistics
Compass : –.
Marácz, László (). Asymmetries in Hungarian. Donostia-San Sebastián: Diputación Foral
de Gipuzkoa.
Marantz, Alec (). On the Nature of Grammatical Relations. Cambridge, MA: MIT Press.
Marantz, Alec (). Clitics, morphological merger, and the mapping to phonological
structure, in Michael Hammond, and Michael Noonan (eds.), Theoretical Morphology.
San Diego, CA: Academic Press, –.
Marantz, Alec (). Cat as a phrasal idiom. MS, Massachusetts Institute of Technology.
Marantz, Alec (). No escape from syntax: Don’t try morphological analysis in the privacy
of your own lexicon. Penn Linguistics Colloquium : –.
References
Marantz, Alec (). Words. Paper presented at the West Coast Conference on Formal
Linguistics (WCCFL) , University of Southern California, February –.
Marantz, Alec (). Subjects and objects. MS, Massachusetts Institute of Technology.
Marantz, Alec (). Objects out of the lexicon: Objects as events. MS, Massachusetts
Institute of Technology.
Marantz, Alec (). Words and phases, in Sook-Hee Choe (ed.), Phases in the Theory of
Grammar. Seoul: Dong In, –.
Marantz, Alec (). Roots, re- and affected agents: Can roots pull the agent under little v?.
Paper presented at the workshop Roots: Word Formation from the Perspective of ‘Core
Lexical Elements’, Stuttgart, – June.
Marantz, Alec (). Syntactic approaches to argument structure without incorporation.
Paper presented at the workshop Structuring the Argument: A Multidisciplinary Workshop
on the Mental Representation of Verbal Argument Structure, Paris, – September.
Marantz, Alec (). Locality domains for contextual allomorphy across the interfaces, in Ora
Matushansky and Alec Marantz (eds.), Distributed Morphology Today: Morphemes for
Morris Halle. Cambridge, MA: MIT Press, –.
Marchand, Hans (). The Categories and Types of Present-Day English Word-Formation.
Munich: Beck.
Massam, Diane (). Cognate objects as thematic objects, Canadian Journal of Linguistics
: –.
Mateu, Jaume (a). Small Clause Results Revisited, ZAS Papers in Linguistics :
–.
Mateu, Jaume (b). Unselected Objects, in Nicole Dehé and Anja Wanner (eds.), Structural
Aspects of Semantically Complex Verbs. Berlin: Peter Lang, –.
Mateu, Jaume (c). Lexicalization patterns and the locative alternation. MS, Universitat
Autònoma de Barcelona.
Mateu, Jaume (). Argument Structure. Relational Construal at the Syntax-Semantics
Interface. PhD thesis, Universitat Autònoma de Barcelona.
Mateu, Jaume (). Parametrizing verbal polysemy: The case of bake revisited. MS, Uni-
versitat Autònoma de Barcelona. [Published as Mateu, Jaume (). Parametrizing verbal
polysemy: The case of bake revisited. Chicago Linguistics Society : –.]
Mateu, Jaume (). Arguing our way to the Direct Object Restriction on English resultatives,
Journal of Comparative Germanic Linguistics : –.
Mateu, Jaume (a). Argument structure and denominal verbs. MS, Universitat Autònoma
de Barcelona.
Mateu, Jaume (b). On the l-syntax of directionality/resultativity: The case of Germanic
preverbs, in Anna Asbury, Jakub Dotlačil, Berit Gehrke, and Rick Nouwen (eds.), Syntax
and Semantics of Spatial P. Amsterdam: John Benjamins, –.
Mateu, Jaume (). Conflation and incorporation processes in resultative constructions, in
Violeta Demonte and Louise McNally (eds.), Telicity, Change, and State: A Cross-Categorial
View of Event Structure. Oxford: Oxford University Press, –.
Mateu, Jaume and Víctor Acedo-Matellán (). The manner/result complementarity
revisited: a syntactic approach, in María Cristina Cuervo and Yves Roberge (eds.), The
End of Argument Structure. Bingley: Emerald Group Publishing Limited, –.
References
Mateu, Jaume and Laia Amadas (). Syntactic tools for semantic construal. Paper presented
at the st Conference on Tools in Linguistic Theory. Utrecht Institute of Linguistics OTS,
Utrecht, July –.
Mateu, Jaume and Gemma Rigau (). A minimalist account of conflation processes:
Parametric variation at the lexicon-syntax interface, in Artemis Alexiadou (ed.), Theoretical
Approaches to Universals. Amsterdam: John Benjamins, –.
Mateu, Jaume and Gemma Rigau (). Verb-particle constructions in Romance: A lexical-
syntactic account, Probus : –.
Matushansky, Ora (). Head movement in linguistic theory, Linguistic Inquiry : –.
Meillet, Antoine (). De l’expression de l’aoriste en latin, Revue de philologie, de littérature
et d’histoire ancienne : –.
Meillet, Antoine and Joseph Vendryès (). Traité de grammaire comparée des langues
classiques. Paris: Honoré Champion.
Mendikoetxea, Amaya (). Construcciones con se: medias, pasivas e impersonales, in
Ignacio Bosque and Violeta Demonte (eds.), Gramática descriptiva de la lengua española,
vol. : Las construccions sintácticas fundamentales. Relaciones temporales, aspectuales y
modales. Madrid: Real Academia Española and Espasa Calpe, –.
Merchant, Jason (). How much context is enough? Two cases of span-conditioned stem
allomorphy, Linguistic Inquiry (): –.
Meurant, Alain (). Itinera Electronica. Lovaine: Université catholique de Louvain.
Miller, D. Gary (). Complex Verb Formation. Amsterdam: John Benjamins.
Molinari, Danielle (). Les préverbés de facio, in Claude Moussy (ed.), La composition et la
préverbation en latin. Paris: Presses de l’Université Paris-Sorbonne, –.
Moussy, Claude (). La polysémie du préverbe com-, in Claude Moussy (ed.), La compos-
ition et la préverbation en latin. Paris: Presses de l’Université Paris-Sorbonne, –.
Mulder, Hotze (). Object continuity and elliptical objects in Latin, Romance Linguistics
and Literature Review : –.
Mulder, René (). The Aspectual Nature of Syntactic Complementarion. PhD thesis, Leiden
Universiteit.
Munaro, Nicola (). Alcuni casi di alternanza di struttura argomentale in inglese, in
Gianluigi Borgato (ed.), Teoria del linguaggio e analisi linguistica. XX Incontro di Gramma-
tica Generativa. Padua: Unipress, –.
Napoli, Donna Jo (). Secondary resultative predicates in Italian, Journal of Linguistics :
–.
Narasimhan, Bhuvana, Vittorio di Tomaso, and Cornelia Verspoor (). Unaccusative or
unergative? Verbs of manner of motion, Quaderni del Laboratorio di Linguistica della Scuola
Normale Superiore di Pisa : –.
Nedjalkov, Vladimir P. (ed.) (). Typology of Resultative Constructions. Amsterdam: John
Benjamins.
Neeleman, Ad (). Complex predicates. Utrecht: Onderzoekinstituut voor Taal en Sprak.
Neeleman, Ad and Hans van de Koot (). Bare resultatives, Journal of Comparative
Germanic Linguistics : –.
Neeleman, Ad and Fred Weerman (). The balance between syntax and morphology:
Dutch particles and resultatives, Natural Language & Linguistic Theory : –.
References
Nespor, Marina and Irene Vogel (). Prosodic Phonology. Dordrecht: Foris Publications.
Newell, Heather (). Aspects of the Morphology and Phonology of Phases. PhD thesis,
McGill University.
Nisard, Désiré (ed.) (). Les agronomes latins. Paris: Firmin Didot, Frères, Libraires.
Olsen, Susan (). Der Dativ bei Partikelverben, in Christa Dürscheid, Karl-Heinz Ramers,
and Monika Schwarz (eds.), Festschrift für Heinz Vater. Tübingen, Niemeyer, –.
Oltra-Massuet, Isabel (). On the constituency structure of the Catalan verbs, MIT Working
Papers in Linguistics : –.
Oltra-Massuet, Isabel and Karlos Arregi (). Stress-by-structure in Spanish, Linguistic
Inquiry : –.
Oniga, Renato (). Composition et préverbation en latin: problèmes de typologie, in Claude
Moussy (ed.), La Composition et la préverbation en latin. Paris: Presses de l’Université Paris-
Sorbonne, –.
Oostendorp, Marc van (). Stress as a proclitic in Modern Greek, Lingua : –.
Oya, Toshiaki (). Ground arguments in German particle verbs: A comparison with Dutch
and English, Journal of Germanic Linguistics : –.
Padrosa-Trias, Susanna (). Complex Word-Formation and the Morphology-Syntax Inter-
face. PhD thesis, Universitat Autònoma de Barcelona.
Padrosa-Trias, Susanna and Angelina Markova (). Some remarks on prefixation: Evidence
from Bulgarian, Catalan and English, Interlingüística : –.
Panhius, Dirk (). Gapping in Latin, The Classical Journal : –.
Pantcheva, Marina B. (). Decomposing Path: The Nanosyntax of Directional Expressions.
PhD thesis, Universitetet i Tromsø.
Pantelia, Maria (). Thesaurus Linguae Graecae. Irvine, CA: University of California.
Perrot, Jean (). Préverbes et cas en hongrois, in André Rousseau (ed.), Les Préverbes dans
les langues d’Europe. Introduction à l’étude de la préverbation. Lille: Presses Universitaires du
Septentrion, –.
Pesetsky, David (). Paths and Categories. PhD thesis, Massachusetts Institute of Technology.
Pesetsky, David (). Morphology and logical form, Linguistic Inquiry : –.
Peyraube, Alain (). A diachronic study of directional complements, in Maya Hickmann
and Stéphane Robert (eds.), Space in Languages: Linguistic Systems and Cognitive Categories.
Amsterdam: John Benjamins, –.
Pfau, Roland (). Grammar as Processor: A Distributed Morphology Account of Spontan-
eous Speech Errors. Amsterdam: John Benjamins.
Pfister, Raimund (). Kompetenz in der lateinischen Syntax, in Harm Pinkster (ed.), Latin
Linguistics and Linguistic Theory. Proceedings of the st International Colloquium on Latin
Linguistics. Amsterdam, April . Amsterdam: John Benjamins, –.
Pinault, Georges-Jean (). Le problème du préverbe en indo-européen, in André Rousseau
(ed.), Les Préverbes dans les langues d’Europe. Introduction à l’étude de la préverbation. Lille:
Presses Universitaires du Septentrion, –.
Pinker, Steven (). Learnability and Cognition. Cambridge, MA: MIT Press.
Pinkster, Harm (). On Latin Adverbs. Amsterdam: North-Holland.
Pinkster, Harm (). Tempus, aspect and Aktionsart in Latin, Aufstieg und Niedergang der
Römischen Welt : –.
References
Pinkster, Harm (). Non-accusative second arguments of two-place verbs in Latin, Cua-
dernos de Filologia Clásica : –.
Pinkster, Harm (). Sintaxis y semántica del latín. Madrid: Ediciones Clásicas.
Pottier, Bernard (). Systématique des éléments de relation. Paris: Klincksieck.
Pourcel, Stéphanie and Anetta Kopecka (). Motion events in French: Typological intrica-
cies. MS, University of Sussex and Max Planck Institute for Psycholinguistics.
Puskás, Genoveva (). Word Order in Hungarian: The Syntax of A’-positions. Amsterdam:
John Benjamins.
Pustejovsky, James (). The Generative Lexicon. Cambridge, MA: MIT Press.
Pustejovsky, James (). Generativity and Explanation in Semantics. A Reply to Fodor and
Lepore, Linguistic Inquiry : –.
Pylkkänen, Liina (). Introducing Arguments. Cambridge, MA: MIT Press.
Ramchand, Gillian C. (). First phase syntax. MS, Oxford University. [Later published as
Ramchand ().]
Ramchand, Gillian C. (). Time and the event: The semantics of Russian prefixes, Nordlyd
: –.
Ramchand, Gillian C. (). Verb Meaning and the Lexicon. A First Phase Syntax. Cambridge:
Cambridge University Press.
Randall, Janet, Angeliek van Hout, Jürgen Weissenborn, and Harald Baayen (). Acquiring
unaccusativity: A cross-linguistic look, in Artemis Alexiadou, Elena Anagnostopoulou, and
Martin Everaert (eds.), The Unaccusativity Puzzle. Explorations of the Syntax-Lexicon
Interface. Oxford: Oxford University Press, –.
Rapoport, Tova R. (). Verbs in depictives and resultatives, in James Pustejovsky (ed.),
Semantics and the Lexicon. Dordrecht: Kluwer, –.
Rapoport, Tova R. (). Structure, aspect, and the predicate, Language : –.
Rappaport, Malka and Beth Levin (). What to do with Theta Roles, in Wendy
Wilkins (ed.), Syntax and Semantics : Thematic Relations. San Diego, CA: Academic
Press, –.
Rappaport Hovav, Malka and Beth Levin (). Building verb meanings, in Miriam Butt and
Wilhelm Geuder (eds.), The Projection of Arguments: Lexical and Syntactic Constraints.
Stanford, CA: CSLI, –.
Rappaport Hovav, Malka and Beth Levin (). An event structure account of English
resultatives, Language : –.
Rappaport Hovav, Malka and Beth Levin (). Reflections on Manner/Result Complemen-
tarity, in Edit Doron, Malka Rappaport Hovav, and Ivy Sichel (eds.), Syntax, Lexical
Semantics, and Event Structure. Oxford: Oxford University Press, –.
Real Puigdollers, Cristina (). La classe dels verbs de mesura: Un cas de verbs fronterers.
Master’s thesis, Universitat Autònoma de Barcelona.
Real Puigdollers, Cristina (). A microparametric approach on goal of motion construc-
tions: Properties of adpositional systems in Romance and Germanic, Catalan Journal of
Linguistics : –.
Real Puigdollers, Cristina (). Lexicalization by Phase. The Role of Prepositions in Argument
Structure and its Cross-linguistic Variation. PhD thesis, Universitat Autònoma de Barcelona.
References
Reinhart, Tanya (). The Theta System: Syntactic realization of verbal concepts. Utrecht:
UiL OTS Working Papers.
Reinhart, Tanya (). The theta system: An overview. Theoretical Linguistics : –.
Reinhold, Heinz (). Zum lateinischen Verbalaspekt, Zeitschrift für Vergleichende Sprach-
forschung auf dem Gebiete der Indogermanischen Sprachen : –.
Richards, Marc D. (). On feature inheritance: An argument from the Phase Impenetra-
bility Condition, Linguistic Inquiry : –.
Rigau, Gemma (). Locative sentences and related constructions in Catalan: ésser/haver
alternation, in Amaya Mendikoetxea and Myriam Uribe-Etxebarria (eds.), Theoretical Issues
at the Morphology-Syntax Interface. Bilbao and Donostia-San Sebastián: Universidad del
País Vasco and Diputación Foral de Gipuzkoa, –.
Ritter, Elizabeth and Sarah T. Rosen (). Delimiting events in syntax, in Miriam Butt and
Wilhelm Geuder (eds.), The Projection of Arguments: Lexical and Syntactic Constraints.
Stanford, CA: CSLI, –.
Rivero, María-Luisa (). Adverb incorporation and the syntax of adverbs in modern Greek,
Linguistics and Philosophy : –.
Roberts, Ian G. (). Agreement and Head Movement: Clitics, Incorporation, and Defective
Goals. Cambridge, MA: MIT Press.
Rojina, Nina (). English Particles, Russian Prefixes, and Prepositional Phrases. Master’s
thesis, Universitetet i Tromsø.
Romagno, Domenica (). Azionalità e transitività: il caso dei preverbi latini, Archivio
Glottologico Italiano : –.
Romagno, Domenica (). Ancora su preverbazione e sistemi verbali. Il caso dei preverbi
greci, Archivio Glottologico Italiano : –.
Romagno, Domenica (). Applicative and causative: Some further reflections upon verbal
prefixation in Greek and Latin, Archivio Glottologico Italiano : –.
Roßdeutscher, Antje (). A syntax-semantics interface for verbs with P-elements in
German, Working Papers of the SFB Incremental Specification in Context: –.
Rosselló, Joana (). El SV I: Verb i arguments verbals, in Joan Solà, Maria Rosa Lloret, Joan
Mascaró, and Manuel Pérez Saldanya (eds.), Gramàtica del català contemporani, vol. .
Barcelona: Empúries, –
Rubio Fernández, Lisardo and Tomás González Rolán (). Nueva gramática latina. Madrid:
Coloquio Editorial.
Schäfer, Florian (). The Syntax of (Anti-)Causatives. External Arguments in Change-of-
State Contexts. Amsterdam: John Benjamins.
Schütze, Carson T. (). Serbo-Croatian second position clitic placement and the
phonology-syntax interface, MIT Working Papers in Linguistics : –.
Selkirk, Elisabeth (). Phonology and Syntax: The Relation between Sound and Structure.
Cambridge, MA: MIT Press.
Serbat, Guy (). Essai de définition du datif latin, in Alfred Bammesberger., and Friedrich
Heberlein (eds.), Akten des VIII. Internationalen Kolloquiums zur Lateinischen Linguistik.
Proceedings of the Eighth International Colloquium on Latin Linguistics. Heidelberg:
Universitätsverlag C. Winter, –.
References
Serbat, Guy (a). Le rôle syntaxique du préverbe: Étude à partir de gradior, Studia
Philologica Valentina : –.
Serbat, Guy (b). Autour de l’objet interne, in Benjamín García Hernández (ed.), Estudios
de lingüística latina. Actas del IX Coloquio Internacional de Lingüística Latina. Universidad
Autónoma de Madrid, – de Abril de . Madrid: Ediciones Clásicas, –.
Shim, Ji Y. and Marcel den Dikken (). The tense of resultatives: The case of Korean. MS,
City University of New York. [Later published as Shim, Ji Y. and Marcel den Dikken ().
The tense of resultatives: The case of Korean, North-Eastern Linguistic Society : –.]
Siegel, Dorothy (). Topics in English Morphology. PhD thesis, Massachusetts Institute of
Technology.
Simpson, Jane (). Resultatives, in Lori Levin, Malka Rappaport Hovav, and Annie Zaenen
(eds.), Papers in Lexical-Functional Grammar. Bloomington, IN: Indiana University Lin-
guistics Club, –.
Slobin, Dan I. (). The many ways to search for a frog: Linguistic typology and the
expression of motion events, in Sven Strömqvist and Ludo Verhoeven (eds.), Relating Events
in Narrative: Typological and Contextual Perspectives. Mahwah, NJ: Lawrence Erlbaum
Associates, –.
Slobin, Dan I. (). Relating narrative events in translation, in Dorit Ravid and Hava Bat-
Zeev Shyldkrot (eds.), Perspectives on Language and Language Development: Essays in
Honor of Ruth A. Berman. Dordrecht: Kluwer, –.
Slobin, Dan I. (). What makes manner of motion salient?, in Maya Hickmann and
Stéphane Robert (eds.), Space in Languages: Linguistic Systems and Cognitive Categories.
Amsterdam: John Benjamins, –.
Smith, Carlotta (). The Parameter of Aspect. Dordrecht: Kluwer.
Snyder, William (). Language Acquisition and Language Variation: The Role of Morph-
ology. PhD thesis, Massachusetts Institute of Technology.
Snyder, William (). On the nature of syntactic variation: Evidence from complex predi-
cates and complex word-formation, Language : –.
Snyder, William (). Parameter theory and motion predicates, in Violeta Demonte and
Louise McNally (eds.), Telicity, Change, and State: A Cross-Categorial View of Event
Structure. Oxford: Oxford University Press, –.
Son, Minjeong (). Directionality and resultativity: The cross-linguistic correlation
revisited, Nordlyd : –.
Son, Minjeong (). Resultatives in Korean revisited: Complementation versus adjunction,
Nordlyd : –.
Son, Minjeong and Peter Svenonius (). Microparameters of cross-linguistic variation:
Directed motion and resultatives, West Coast Conference on Formal Linguistics : –.
Sorace, Antonella (). Gradients in auxiliary selection with intransitive verbs, Language :
–.
Spencer, Andrew and Marina Zaretskaya (). Verb prefixation in Russian as lexical
subordination, Linguistics : –.
Starke, Michal (). Nanosyntax. A short primer to a new approach to language, Nordlyd
: –.
References
Talmy, Leonard (). Toward a Cognitive Semantics, vol. : Typology and Process in Concept
Structuring. Cambridge, MA: MIT Press.
Tenny, Carol (). Grammaticalizing Aspect and Affectedness. PhD thesis, Massachusetts
Institute of Technology.
Tenny, Carol (). Aspectual Roles and the Syntax-Semantics Interface. Dordrecht: Kluwer.
Terzi, Arhonto (). On null spatial Ps and their arguments, in Guglielmo Cinque and Luigi
Rizzi (eds.), The Cartography of Syntactic Structures, vol. : Mapping Spatial PPs. Oxford:
Oxford University Press, –.
Thayer, William P. (). LacusCurtius. University of Chicago.
Tombeur, Paul (ed.) (). Bibliotheca Teubneriana Latina (CD-ROM). Munich: K. G. Saur
and Turnhout: Brepols.
Tomioka, Naoko (). Resultative Constructions: Cross-Linguistic Variation and the Syntax-
Semantics Interface. PhD thesis, McGill University.
Torrego, Esther (). Unergative-unaccusative alternations in Spanish, MIT Working Papers
in Linguistics : –.
Torrego, Ma Esperanza (). Restricciones de la categoría verbal aspecto sobre los elementos
de duración en latín, Minerva : –.
Tortora, Christina (). Verbs of inherently directed motion are compatible with resultative
phrases. Linguistic Inquiry : –.
Travis, Lisa deMena (). Parameters and Effects of Word Order Variation. PhD thesis,
Massachusetts Institute of Technology.
Travis, Lisa deMena (). Event structure in syntax, in Carol Tenny and James Pustejovsky
(eds.). Events as Grammatical Objects. Stanford, CA: CSLI, –.
Tungseth, Mai E. (). Two structural positions for locative and directional PPs in Norwe-
gian motion constructions, Nordlyd : –.
Turcan, Isabelle (). La dépréverbation dans les langues classiques, Bulletin de la Société de
Linguistique de Paris : –.
Untermann, Jürgen (). Sprachwandel, beobachtet an lateinischen Präverbien, in Hannah
Rosén (ed.), Aspects of Latin: Papers from the Seventh International Colloquium on Latin
Linguistics, Jerusalem, April . Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft,
–.
Vairel, Hélène (). La Valeur de l’opposition infectum/perfectum en latin, Revue des Études
Latines : –.
Van der Heyde, Klaas (). L’aspect verbal en latin. Problèmes et résultats. Le rôle du
préverbe, Revue des études latines : –.
Van Laer, Sophie (). Le Préverbe ob- en latin: Considérations sémantiques, in Benjamín
García Hernández (ed.), Estudios de lingüística latina. Actas del IX Coloquio Internacional de
Lingüística Latina. Universidad Autónoma de Madrid, – de Abril de . Madrid:
Ediciones Clásicas, –.
Van Riemsdijk, Henk C. (). A Case Study in Syntactic Markedness. The Binding Nature of
Prepositional Phrases in Dutch. Lisse: The Peter de Ridder Press.
Van Valin, Robert D. Jr. (). A synopsis of role and reference grammar, in Robert D. Van
Valin Jr. (ed.), Advances in Role and Reference Grammar. Amsterdam: John Benjamins,
–.
References
Vendler, Zeno (). Linguistics in Philosophy. Ithaca, NY: Cornell University Press.
Vendryès, Joseph (). Les Notions de ‘boire’ et ‘manger’ en indo-européen et l’aspect verbal,
Bulletin de la Société de Linguistique : –.
Verkerk, Annemarie (). Diachronic change in Indo-European motion event encoding,
Journal of Historical Linguistics : –.
Verkuyl, Henk J. (). On the Compositional Nature of the Aspects. Reidel: Dordrecht.
Verkuyl, Henk J. (). A Theory of Aspectuality. The Interaction between Temporal and
Atemporal Structure. Cambridge: Cambridge University Press.
Vikner, Sten (). Verb Movement and Expletive Subjects in the Germanic Languages.
Oxford: Oxford University Press
Vincent, Nigel (). The evolution of c-structure: Prepositions and PPs from Indo-European
to Romance, Linguistics : –.
Washio, Ryuichi (). Resultatives, compositionality and language variation, Journal of East
Asian Linguistics : –.
Wechsler, Stephen (). Resultatives under the ‘event-argument homomorphism’ model of
telicity. Paper presented at the workshop The Syntax of Aspect, Ben Gurion University of
Negev, June. [Later published in Nomi Erteschik-Shir and Tova Rapoport (eds.) (), The
Syntax of Aspect. Deriving Thematic and Aspectual Interpretation. Oxford: Oxford Univer-
sity Press, –.]
Wechsler, Stephen (). Resultatives under the ‘event-argument homomorphism’ model of
telicity, in Nomi Erteschik-Shir, and Tova Rapoport (eds.), The Syntax of Aspect. Deriving
Thematic and Aspectual Interpretation. Oxford: Oxford University Press, –.
Weerman, Fred (). The V Conspiracy. A Synchronic and a Diachronic Analysis of Verbal
Positions in Germanic Languages. Dordrecht: Foris.
Whelpton, Matthew (). Resultatives in Icelandic. A preliminary investigation. MS, Uni-
versity of Iceland.
Whelpton, Matthew (). Building resultatives in Icelandic, Western Conference on Linguis-
tics : –.
Williams, Edwin (). Argument structure and morphology, The Linguistic Review : –.
Wunderlich, Dieter (). An investigation of lexical decomposition: The case of German
be-verbs, Linguistics : –.
Wunderlich, Dieter (). Towards a structural typology of verb classes, in Dieter Wunder-
lich (ed.), Advances in the Theory of the Lexicon. The Hague: Mouton de Gruyter, –.
Wurff, Wim van der (). Null objects and learnability: The case of Latin, Working Papers of
Holland Institute for Generative Linguistics : –.
Wurmbrand, Susanne (). The structure(s) of particle verbs. MS, University of
Connecticut.
Xu, Dan (). Typological Change in Chinese Syntax. Oxford: Oxford University Press.
Žaucer, Rok (). The Role of Verbal Prefixes in Slavic: Evidence from Slovenian Locative
Denominal Verbs. Master’s thesis, University of Ottawa.
Žaucer, Rok (). Slavic prefixes as state morphemes. From state to change-of-state and
perfectivity, in Wolfgang U. Dressler, Dieter Kastovsky, Oskar E. Pfeiffer, and Franz Rainer
(eds.), Morphology and its Demarcations. Selected Papers from the th Morphology Meeting,
Vienna, February . Amsterdam: John Benjamins, –.
References
Gaffiot, Félix , , , , , , , Halliday, Michael A. K. n.
, , , Harley, Heidi , –, , , , , ,
galleggiare (Italian verb) –, , – , , n., , , n.,
García Hernández, Benjamín n., , Haudry, Jean
n., , , Haugen, Jason D. n. , n.
Gehrke, Berit , , , , , , n., HAVE-auxiliary, see auxiliary selection
, , , , , , , , Haverling, Gerd , , n., n.
, , Hay, Jennifer n.
generative-constructivist theory ; see also head:
exo-sketal approach; complex , , , –
neo-constructionist approach cyclic head, see cycle: cyclic head
German , , n., , , –, n. , eventive head –, , , –
, , n., n., , n., , , head movement , , , ,
, –, , –, n., , Head Movement Constraint
Germanic , , , n. , , , , , lexical –, –,
, , , , , , –, and Morphological Merger –, ,
gerundive construction n. see also functional head; specifier: specifier-
Ghomeshi, Jila n. head relation; v
Gianollo, Chiara n. Hebrew , , , –
Gibert Sotelo, Elisabeth n. Hegedűs, Veronika
Gleitman, Lila n. Heslin, Thomas P. Jr. n.
goal , Hewson, John n.
in the dative Hoekstra, Teun , , , , , ,
as opposed to probe , , n., , ,
Goldberg, Adele n., n. Hofmann, Johann B. , , , n.,
González Rolán, Tomás –, ,
grammatical formative holistic effect , –,
grammatically irrelevant meaning n. hornear (Spanish verb)
grammatically relevant meaning n., , Horrocks, Geoffrey –, , –,
Greek , , –, n., ,
Green, Georgia Horvath, Julia
Grimshaw, Jane n. Hout, Angeliek van n. , ,
Ground , , –, Hungarian , , , –
Central Ground
promoted I-phrasing
Terminal Ground Iacobini, Claudio –
see also Unselected Object Contruction: Icelandic n. , , –, ,
Ground Unselected Object Contruction idiomatic interpretation ; see also non-
Guo, Jiansheng n. compositional meaning; special
meaning
Haider, Hubert immisceo (Latin verb) –
Hale, Kenneth , –, , , , , , imperfective, see viewpoint aspect:
, , , , imperfective
Halle, Morris , , , , , , , in- (Latin prefix) –
Index
lexical entry , , MacDonald, Jonathan E. , , , ,
lexical gap ,
lexical item –, , – McIntyre, Andrew n. , , , , ,
in Hale and Keyser’s theory , , n. , n. , , ,
lexical marking n., –, n., n., , , , ,
lexical semantics , , malefactive dative, see dative: benefactive/
and deponency malefactive
lexical semantic representation – Mandarin Chinese
lexical subordination , Manner , ; see also Co-event; verb:
lexicalist analysis ; see also endo-skeletal manner-of-motion verb
approach; projectionist theory mapping algorithm ,
lexicalization , , Marácz, László
in Nanosyntax ; see also Exhaustive Marantz, Alec , , , –, n. ,
Lexicalization Principle –, , n., , –, –, ,
lexicon-syntax interface , , , , n., , n.,
LF –, , Marchand, Hans n.
Liddell, Henry G. Markova, Angelina , , , ,
Lieber, Rochelle n. mass , , –, , ; see also
Lindvall, Ann quantity: non-quantity
Linearization , , – Massam, Diane n. ,
linear intervention , –, Mateu, Jaume , n., , , –, , , ,
listeme , , –; see also root , , , , , , , , , , , ,
little a, see a , n., , , , , n.,
little n, see n , n. , , n., n. , ,
little p, see p , n., , , , , –
little v, see v Matushansky, Ora n.
locality , , measure phrase –
Location (theta-role) Measurer –, –, –
Locative Alternation –, – meg (Hungarian particle)
with adjectives n. Meillet, Antoine n., , ,
derivational analysis Mendikoetxea, Amaya n.
non-derivational analysis Merchant, Jason
and prefixation – Merge ,
and the s-/v-framed distinction – metonymy
see also change of location: change-of- Meurant, Alain
location alternant; change of state: microparametric theory, see cross-linguistic
change-of-state alternant; fill variation: as microparametric
locatum verb, see verb: location/locatum verb Miller, D. Gary , , n.
Löfstedt, Bengt Minimalist Program ,
Lohndal, Terje n. minimality
López Moreda, Santiago n. Modern Greek, see Greek
Lowering Molinari, Danielle
Lüdeling, Anke , , morph
Luraghi, Silvia , , portmanteau morph ,
Index
Published InterPhases
The Syntax of Silence Phase-Theoretic Investigations of Linguistic Interfaces
Sluicing, Islands, and the Theory of Ellipsis edited by Kleanthes Grohmann
by Jason Merchant Negation in Gapping
Questions and Answers in Embedded Contexts by Sophie Repp
by Utpal Lahiri A Derivational Syntax for Information Structure
Phonetics, Phonology, and Cognition by Luis López
edited by Jacques Durand and Bernard Laks edited by Anastasia Giannakidou and Monika Rathert
At the Syntax-Pragmatics Interface The Syntax of Sentential Stress
Concept Formation and Verbal Underspecification in by Arsalan Kahnemuyipour
Dynamic Syntax Tense, Aspect, and Indexicality
by Lutz Marten by James Higginbotham
The Unaccusativity Puzzle Lexical Semantics, Syntax, and Event Structure
Explorations of the Syntax-Lexicon Interface edited by Malka Rappaport Hovav, Edit Doron, and Ivy
edited by Artemis Alexiadou, Elena Anagnostopoulou, Sichel
and Martin Everaert
About the Speaker
Beyond Morphology Towards a Syntax of Indexicality
Interface Conditions on Word Formation by Alessandra Giorgi
by Peter Ackema and Ad Neeleman
The Sound Patterns of Syntax
The Logic of Conventional Implicatures edited by Nomi Erteschik-Shir and Lisa Rochman
by Christopher Potts edited by Phoevos Panagiotidis
Paradigms of Phonological Theory Interfaces in Linguistics
edited by Laura Downing, T. Alan Hall, and Renate New Research Perspectives
Raffelsiefen edited by Raffaella Folli and Christiane Ulbrich
The Verbal Complex in Romance Negative Indefinites
by Paola Monachesi by Doris Penka
The Syntax of Aspect Events, Phrases, and Questions
Deriving Thematic and Aspectual Interpretation by Robert Truswell
Edited by Nomi Erteschik-Shir and Tova Rapoport
Dissolving Binding Theory
Aspects of the Theory of Clitics by Johan Rooryck and Guido Vanden Wyngaerd
by Stephen Anderson The Logic of Pronominal Resumption
Canonical Forms in Prosodic Morphology by Ash Asudeh
by Laura J. Downing Modals and Conditionals
Aspect and Reference Time by Angelika Kratzer
by Olga Borik The Theta System
Direct Compositionality Argument Structure at the Interface
edited by Chris Barker and Pauline Jacobson edited by Martin Everaert, Marijana Marelj, and Tal
A Natural History of Infixation Siloni
by Alan C. L. Yu Sluicing
Phi-Theory Cross-Linguistic Perspectives
Phi-Features Across Interfaces and Modules edited by Jason Merchant and Andrew Simpson
edited by Daniel Harbour, David Adger, and Susana Béjar Telicity, Change, and State
French Dislocation A Cross-Categorial View of Event Structure
Interpretation, Syntax, Acquisition edited by Violeta Demonte and Louise McNally
by Cécile De Cat Ways of Structure Building
Inflectional Identity edited by Myriam Uribe-Etxebarria and Vidal Valmala
edited by Asaf Bachrach and Andrew Nevins The Morphology and Phonology of Exponence
Lexical Plurals edited by Jochen Trommer
by Paolo Acquaviva Count and Mass Across Languages
Adjectives and Adverbs edited by Diane Massam
Syntax, Semantics, and Discourse Genericity
Edited by Louise McNally and Christopher Kennedy edited by Alda Mari, Claire Beyssade, and Fabio Del Prete
Strategies of Quantification by Daniela Isac
edited by Kook-Hee Gil, Steve Harlow, and George Sentence and Discourse
Tsoulas edited by Jacqueline Guéron
Nonverbal Predication Optimality-Theoretic Syntax, Semantics, and
Copular Sentences at the Syntax-Semantics Interface Pragmatics
by Isabelle Roy From Uni- to Bidirectional Optimization
Diagnosing Syntax edited by Géraldine Legendre, Michael T. Putnam, Henri-
edited by Lisa Lai-Shen Cheng and Norbert Corver ëtte de Swart, and Erin Zaroukian
Pseudogapping and Ellipsis The Morphosyntax of Transitions
by Kirsten Gengel A Case Study in Latin and Other Languages
Syntax and its Limits Víctor Acedo-Matellán
edited by Raffaella Folli, Christina Sevdali, and Robert
Truswell Published in association with the series
The Oxford Handbook of Linguistic Interfaces
Phrase Structure and Argument Structure
edited by Gillian Ramchand and Charles Reiss
A Case Study of the Syntax-Semantics Interface
by Terje Lohndal In preparation
Edges in Syntax Modality Across Syntactic Categories
Scrambling and Cyclic Linearization edited by Ana Arregui, Mari Luisa Rivero, and Andrés
by Heejeong Ko Pablo Salanova
The Syntax of Roots and the Roots of Syntax Phi Syntax
edited by Artemis Alexiadou, Hagit Borer, and Florian A Theory of Agreement
Schäfer by Susana Béjar
Causation in Grammatical Structures Stratal Optimality Theory
edited by Bridget Copley and Fabienne Martin by Ricardo Bermúdez Otero
Continuations and Natural Language Phonology in Phonetics
by Chris Barker and Chung-chieh Shan by Abigail Cohn
The Semantics of Evaluativity Concealed Questions
by Jessica Rett by Ilaria Frana
External Arguments in Transitivity Alternations Lexical Semantics and Morphosyntactic Variation
by Artemis Alexiadou, Elena Anagnostopoulou, and Flor- by Itamar Francez and Andrew Koontz-Garboden
ian Schäfer Generality and Exception
Control and Restructuring by Ivan Garcia-Alvarez
by Thomas Grano Computing Optimality
The Interaction of Focus, Givenness, and Prosody by Jason Riggle
A Study of Italian Clause Structure Pragmatic Aspects of Scalar Modifiers
The Morphosyntax of Gender by Osamu Sawada
by Ruth Kramer Gradience in Split Intransitivity
The Morphosyntax of Imperatives by Antonella Sorace