Locative and relator Ps in Romance

Locative and relator Ps in Romance
M. Rita Manzini (Università degli Studi di Firenze) – Ludovico Franco (CLUNL / Universidade
Nova de Lisboa) – Leonardo M. Savoia (Università di Firenze)
The issue. Location in languages is often externalized by the same cases/Ps introducing genitive,
dative, instrumental, i.e. non-spatial obliques (Caha 2009). There are two possible ways of
accounting for this coincidence in the syntax (avoiding the conclusion that a mere morphological
syncretism is involved): either the spatial meaning is primitive and the relational meaning is derived
– or vice versa. We provide evidence for the second approach.
Background and proposal. We start from Manzini and Savoia (2011 ff.), who propose that Italian
a ‘to’ is a part-whole or inclusion predicate, notated (⊆), whose basic context of occurrence is the
dative, as illustrated for English ‘to’ in (1).
(1)
a.
I give the book to Peter
b.
give [PredP the book [[⊆ to] Peter ]]]
As a consequence, Italian a (English to etc.) will be interpreted as a locative only in so far as it
denotes locatively constrained inclusion (⊆). In other words, locative is a specialization of the
dative relation, which arises from the locative nature of the element embedded by a/to (cf. the
difference between ‘I sent the letter to Peter’ and ‘I sent the letter to Rome’), or from the locative
nature of the verb a/to attaches to, or other. We argue in favour of a derived status for locatives, by
showing that locative properties, such as motion vs. state or directionality, cannot be part of the core
meaning of Ps like a etc. Rather these Ps are general relators, and the locative interpretation
depends on their context of embedding.
Case study 1: Locative DOM. Italian a ‘at/to’ introducea both state-in and motion-to, as in (2a),
while da ‘from’ specializes for motion-from, as in (2b). However, this is true only of inanimate
complements. State-in, motion-to and motion-from with human referents are all externalized by da
(for which the ‘from’ translation turns out to be inaccurate), as in (3). This configures an instance of
Differential Object Marking (DOM) in the P domain (cf. Fabregas 2015). From (3), we conclude
that directionality and other spatially salient specifications (state/motion) are missing from da’s
core denotation; if they were present, its compatibility with the different contexts in (3) could not be
explained.
(2)
a.
Sono/vado
a/*da casa
b.
Esco
*a/da
casa.
I.am/I.go
at/to/*from home
I.get.out
*at/*to/from home
(3)
Sono/vado/esco
da-l
parrucchiere.
I.am/I.go/I.get.out
at/to/from-the hairdresser
‘I am at/I go to/I come from the hairdresser’
Italian
By hypothesis, a ‘at/to’ in (2) has the content ⊆, meaning roughly that ‘house’ locatively contains
(or comes to contain) ‘me’. The same ⊆ content should characterize da in (3), which subsumes the
same contexts. One piece of evidence that this conclusion is on the right track comes from the fact
that at least motion-from da is pronominalized by the same clitic that pronominalizes
genitive/partitives, namely ne, as in (4) (motion-to and state-in have a dedicated clitic ci
‘there/here’, syncretic with the dative clitic in many Italian dialects, cf. Manzini and Savoia 2007,
Kayne 2010).
(4)
a.
Ne
esco ora
(dal parrucchiere/da casa)
from.it I.get.out now (from the hairdresser/ home)
b.
Ne
vedo tre
of.them I.see three
‘I see three of them’
This leaves open the question as to how we account for the meaning differences triggered by a vs
da, given their common interpretive content, as well as the question regarding the DOM(-like) split.
We return to these issues after introducing more evidence against a/da being intrinsically locative.
Case Study 2: Umbria dialects. In the variety in (5), da includes the contexts reviewed for Italian
(2)-(3), namely motion-from (5c) and locative DOM (5b); at the same time da also introduces
dative complements (5a). a is also present in this language and introduces motion-to in (5d).
(5)
a.
lo
ðajo da
mmi fratɛllo
it
I.give to
my
brother
b.
vajo/stɔnno lla
da
esso
I.go/they.stay there to/at him
c.
vɛŋgo da
su/ju
kkasa
I.come from up/down
home
d.
vajo a/*da ttɔdi
I.go to
Todi
Avigliano (Umbria)
Face to a system like Italian (2) (or face to English) one may entertain the idea that the denotation
of goals in ditransitive constructions is an extension of the motion-to content. However (5) is a
direct counterexample to this, since the goal can equally well be encoded by the same P that is
found in motion-from contexts. The simplest analysis (supported by the occurrence of da with
highly-ranked objects) is that Ps introduce elementary relations that do not pertain to the locative
domain; the locative meaning depends on extra locative restrictions on their relational core.
Other Romance languages. Though Italian varieties have the da P, many Romance languages,
including in particular French, introduce motion-from with the genitive P de ‘of’ as in (6a-b); again
a dedicated P introduces highly ranked referents, namely chez, as in (6c). The important point is that
French de ‘from’ is again taken from the basic repertory of relators of the language. In fact genitive,
like dative, denotes possession/inclusion/part-whole and is therefore assigned the ⊆ content in the
present work. Therefore, though the contours of the system are slightly redrawn in French, they
remain consistent with the conclusions reached in connection with Italian.
(6)
a.
Je viens
de la maison/ de Paris
I come
from the house/from Paris
b.
le livre
de Marie
the book
of Marie
c.
Je suis/vais chez lui
I am/go
at/to him
Differentiating state, motion and direction. Assuming that our line of reasoning is on the right
track we need a way to formally represent the different locative interpretations of a and da in Italian
(a and da in Aviglianese, à and de in French, etc.). Consider Italian. As stated at the outset, location
can be conceived as inclusion in location and it is therefore sufficient to stipulate a locative
restriction on ⊆ to capture it, as in (7). However this does not differentiate a and da.
(7)
a, da: P, ⊆(Loc)
It seems to us that the only logical possibility open is that the relevant interpretations depend on the
shape of the event. In telic events, a ‘to’ expresses location at the telos of the event, i.e. at the result
clause implied by it (Higginbotham 2009), as in ‘I go (to) home’ in (2a) with the structure in (8a)
(see Alexiadou et al. 2015 for unaccusatives as verbs endowed with a v layer). By contrast, da
‘from’ implies location outside the result or telos of the event, hence at its causal component, as in
‘I get out from home’ in (8b), cf. (2b). This conclusion is confirmed by the fact that English from
also introduces causers, while Italian da introduces causers and agents (i.e. ‘by’-phrases).
(8)
a.
[CAUSE
[ResP go [⊆ to home]]]
b.
[CAUSE
[ResP get out] [⊆ from home]]
State-in, as the most basic locative relation, requires only that P embeds a location, as in (9), where
the atelic predicate blocks the motion-to reading in Romance, but not the state-in reading.
(9)
Ho camminato a Roma
I walked in/*to Rome
Prepositional DOM. As for locative DOM, it is not difficult to account for languages like French
where Ps such as chez simply specialize for locative inclusion with highly ranked referents. The
same choice of a specialized locative P can be seen in many Italian dialects, e.g. nnə/ndə (Aidone,
Sicily, Manzini and Savoia 2005, etymologically related to Latin inde ‘from there’). One possible
way of approaching Italian da is to assume that highly ranked referents can never be embedded
under a low locative as in (8b), but must be embedded at least as high as in (8a), independently of
directionality.