Abstract-ralfe2012

French anticausatives do not differ in meaning but they compete
1. Verbs undergoing the causative/anticausative alternation divide into two morphological and
three distributional classes in French: With verbs of class A, the anticausative (AC) is
morphologically unmarked (∅-ACs), with verbs of class B, the AC is marked with the
reflexive clitic se (se-ACs), and ACs of class C allow both markings (∅/se-ACs). Some
authors have proposed that the presence vs. absence of the reflexive clitic goes along with
subtle differences in meaning. To derive these, fundamentally different syntactic structures
have been proposed for se-ACs and ∅-ACs. For example, Labelle 1992 argues that se-ACs are
unaccusative while ∅-ACs are unergative. Doron & Labelle 2010 (henceforth DL) propose
that se-ACs and ∅-ACs are both unaccusative but differ substantially in their event
decomposition and the position where the lexical root is merged in the structure. In this paper,
we show that most alleged meaning differences between se-ACs and ∅-ACs are either not
existent or idiosyncratic/verb-specific. To the extent that meaning aspects can be robustly
associated with either marked or unmarked ACs, this holds only for verbs of class C but does
not generalize to morphologically identical ACs in class A or B. This makes a structural
explanation of these meaning differences unfeasible: the presence vs. absence of se cannot be
associated with syntactic differences driving meaning differences. To explain the remaining
robust differences in interpretation and distribution and why they occur only with ACs of
class C, we propose a pragmatic explanation. With verbs of class C, a pragmatic reasoning on
the possible interpretations of the string [DP se V] (AC or also semantically reflexive) leads
the speaker to prefer one version over the other. Note that we do not deny any syntactic
differences between ∅-ACs and se-ACs: The presence of se suggests a syntactic extra-layer
on top of VP, a middle or expletive Voice (Doron 2003, Alexiadou et al. 2006). This projec-
tion triggers syntactic differences (e.g. auxiliary selection) but does not add any semantics.
2. It was claimed that the change of state (COS) is presented as internally caused when
expressed by ∅-ACs and as externally caused when expressed by se-ACs (C1, Rothemberg
1974, Labelle 92, DL). This claim explains the alleged contrast in (1a-b), but it is problematic.
(1) a. Il vit le mouchoir *(se) rougir
He saw the handkerchief REFL redden First, descriptions in traditional grammars convey exactly the opposite intuition: se-ACs are
said to underline some responsibility of the subject for the COS to evolve (Vendryes 1948,
Gougenheim 1962, Grevisse 1969). We backed up this intuition with a small experiment: 8
native speakers were asked to pick among the se-AC and ∅-AC of 4 verbs of class C the one
which attributes more responsibility to the subject (inanimate in all examples) and they all
chose the se-AC. Second, when the COS is internally caused, only verbs of class C with
animate subjects are problematic as se-AC. While (1b) and (2a, b) are bad with se-ACs, the
same verbs form good se-ACs in (3a, b). (4a, b) show that the COS in (3a, b) is conceived as
internally caused, since the verb+object combinations at hand do not transitivize (Levin &
Rappaport 1995). Verbs of class B allow an AC-reading with animate subjects (cp. 5 with 2b).
(2) a. Pierre *s'est/a vieilli (Pierre grew older) b. Pierre *s'est/a maigri (Pierre thinned)
(3) a. Ses yeux se sont rougis à l'annonce de la victoire
Her eyes REFL are reddened at the announcement of the victory
b. Le vin est un produit qui vit, [.] se vieillit (The wine is a product that lives, REFL ages)
(4) a. *L'annonce de la victoire a rougi ses yeux b. *Le producteur/stockage a vieilli le vin
The announcement of victory has redden her eyes The producer/storing aged the wine
(5) Marie s'est amaigrie/??a amaigri (Marie REFL is a-thinned)
Third, when the COS is externally caused, the ∅-AC is not systematically ungrammatical,
contra C1. For us, (1a) is perfect without se, and we find examples such as in (6) on the web.
(6) [Son] cou a rougi au soleil (Her neck has reddened from the sun)
Fourth, verbs of class B can express internally caused events (L'univers s'agrandit).
3. As a further argument for C1, Labelle 1992 and DL present the fact that se-ACs (both from
class B and C) are excluded from the construction mettre x à P, cf. (7). DL assume that this
construction describes "the fact of creating the appropriate conditions for an autonomous
process to take place." Given C1, se-ACs are predicted to be excluded from the structure.
(7) Il a mis le sucre à (*se) caraméliser (He has put the sugar to (REFL) caramelize)
This argumentation misses the fact that transitive verbs (not expressing a COS or expressing
an externally caused COS) enter the structure (8a) while clearly internally caused COS verbs
are out (8b) and that verbs of class B enter the construction if the reflexive is dropped (8c).
(8) Il a mis a) le pantalon à laver/à raccourcir b) *les roses à fleurir c) le linge à essorer
He put the trousers to wash/to shorten the roses to blossom the laundry to wring
In our view, mettre x à P is a kind of serial-verb construction. Mettre introduces a subject
acting as direct or indirect causer and selects for a bare VP with an internal argument. This
can be the VP of a transitive verb stripped of its VoiceP (note that passives and unergatives
are out as complements of mettre) or the VP of an AC. As with transitives, se-AC can enter
the construction only if their expletive/middle VoiceP realized as se is cut off.
4. ∅-ACs and se-ACs have also been claimed to show aspectual differences (C2) in that ∅-
ACs focus on the process, whereas se-ACs focus on the final state of the COS. Again, this
lead to the proposal that se-ACs and ∅-ACs differ syntactically (DL, Labelle 1992). However,
there are no convincing arguments for C2 even within ACs of class C. According to Labelle
1992 for-phrases are supposed to be acceptable only with ∅-ACs (9a). However, for us (9a) is
acceptable with both ACs and confirming examples can be found in corpora, cf. e.g. (9b).
(9) a. Le ciment a/*?s'est durci pendant 3 heures (The cement hardened for 3 hours)
b. Le temps s'est radouci pendant 3 ou 4 jours (The weather softened for 3 or 4 days)
The second argument bears on the verb muer, which forms a se-AC only in presence of a
resultative en-PP (10a), while it is claimed to form a ∅-AC only without it (10b). DL argue
that since the se-AC focuses on a result state, an explicit description of this state is needed in
(10a), and since the ∅-AC focuses on the process, the state is de-emphasized and this makes
an explicit state description impossible. We do not think that the paradigm of muer supports
C2. First, we do not agree with the judgement in (10b); cf. also the attested example in (11).
(10) a. L'oiseau s'est mué *(en un monstre à trois têtes) The bird is REFL/has moulted
b. L'oiseau a mué (*en un monstre à trois têtes) (in a monster with three heads)
(11) Le [restaurant] Lézard a mué en un hybride (The restaurant Lézard changed in a hybrid)
Second, while we think that muer with en-PP is an AC (of class C), it is an unergative if it
appears without en-PP (cf. also Zribi-Hertz 1989). We note that (i) only the former can be
transitivized (ii) only the former allows an adjectival passive (l'oiseau est mué *(en monstre))
and (iii) the two are lexicalized differently in other languages (e.g. transform vs. molt).
5. There remain two robust differences between se-ACs and ∅-ACs, but they emerge only
with ACs of class C. i) Human DPs are out in se-ACs (1b)-(2a, b). (ii) The DP is ascribed
more responsibility for the COS in the se-variant than in the ∅-variant (§ below 1a, b). We
provide a pragmatic and competition-driven explanation for these differences which builds
on the fact that the morphological string of a se-AC can be related to two different syntactico-
semantic derivations, either as an inchoative (John became red) or as a semantically reflexi-
vized causative (John reddened himself). Reflexive interpretations typically involve a human
external argument. Therefore, strings as (1b)-(2a, b) are strongly compatible with a reflexive
interpretation and speakers avoid this ambiguity/misinterpretation by shifting, if possible, to
the ∅-variant. Non-human antecedents are rare as external arguments binding a reflexive clitic
and, therefore, a disambiguation is not necessary. But if asked which version of an AC
ascribes more responsibility to the nonhuman DP, speakers will opt for the version that, at
least formally, allows a derivation where the DP is an external argument. Note that verbal and
se-passives are subject to a related competition so that se-passives strongly disfavor human
undergoers (Zribi-Hertz 1982). Finally, while other languages often have only very view ACs
of type C (e.g. German, Italian), these nevertheless show the same interpretational effects.

Source: http://www.umr7023.cnrs.fr/sites/sfl/IMG/pdf/ralfe2012_submission_15.pdf