Deconstructing the conative alternation in Romance

MANGIALAVORI RASIA, MARÍA EUGENIA; AUSENSI, JOSEP

Georgia

Simposio; LSRL 49; 2019

University of Georgia Department of Linguistics

(7) Deconstructing the conative alternation in Romance Josep Ausensi (Universitat Pompeu Fabra) & M. Eugenia Mangialavori Rasia (CONICET) In the conative alternation [CA], the object of the transitive variant is expressed in a prepositional phrase, with a slight change of meaning (telicity/resultativity drop). This construction is attested with verbs whose meaning includes both movement and contact (Guerssel et al., 1985; Goldberg 1995, Levin &Rappaport[LR] 2015, Leek 1996 i.m.a). Even if the CA is fully productive in many languages, like English, Danish, German (Beavers 2006), it is generally claimed to be unavailable in Romance (but see Míguez 2016 on Galician). (1) illustrates the alleged restriction in Spanish. a. Juan golpeó los hombres. b. *Juan golpeó en los hombres. ?Juan hit the men? ?Juan hit (*at) the men? Conversely, drawing on corpus data (12th-17th c., from Sánchez-Marco et al. 2009 [SM] and Corpus del Español [CES]) from Old Spanish [OSp], we show that the CA is possible and productive in Romance grammars. We show that: (i)OSp CA resembles the English alternation in that it allows similar verb classes?cutting (cortar ?cut?, morder ?bite?) (2), contact (dar ?hit?, tallar ?carve?, disparar ?shoot?)(7); and consumption verbs (beber ?drink?,comer ?eat?)(5)?to alternate. (ii) OSp is more flexible, allowing change-of-state/location (COSL) verbs (3) (matar ?kill?(6), destrozar ?shatter?, destruir ?destroy?, derribar ?knock down?, ferir ?hurt?, romper ?break?) and pure contact verbs (4) in CA. In principle, pure contact and COS verbs are not expected in this construction as they lack the required (motion+contact) components (*Janet broke/touched at the vase, Levin 1993, Beavers 2011). Further, the fact that COSL verbs in OSp may alternate in a CA is of particular importance as it suggests the thus far unattested possibility that certain verbs, like break, may enter the causative/inchoative alternation and the CA, which is a pattern often assumed not to occur in natural languages (Levin 1993, 2017 i.a.). a. Y comienço a cenar y morder en mis tripas. (16th c., SM) and started to dine and bite in my guts ?And [I] started dinner and bite at my intestines? Pues tiene vuestra alteza primero que cortar en mi cabeza. (17th c., CES) then have your highness first that cut in my head ?Then your highness has to cut at my head first? a. & mato en ellos fasta que llego a las espannas. (13th c., SM) and killed in them until that arrived at the Spain.PL ?And [he] killed at them until he reached the Spanish realms? Arrebatadamente destroçan en los griegos por duros. (15th c., SM) violently shatter.3P in the Greeks for hard ?They violently shatter at the Greeks because they are strong? Aquel emperador yendo destruyendo en aquella yent. (13th c., SM) that emperor going destroying in that people ?That emperor was destroying at that people? suares fue derribando enellos fasta en la meatad de la puente. (14th c., SM) Suares went knocking-down in-them until in the middle of the bridge ?Suarez advanced knocking down at them up to the middle of the bridge? No se atrevió a romper en ellos, porque los vio fuertes. (16th c., SM) no SE dared to break in them because ACC.3M.PL saw strong ?He didn?t dare to break at them apart, since he saw they were strong? El rey de Portugal prohibio que tocassen en las rayzes. (16th c., SM) the king of Portugal forbade that touch-SBJ-3P in the roots ?The King of Portugal forbade them to touch at the roots.? PROPOSAL: (A) DATA. In CONSUMPTION VERBS, OSp data seems to support the observation that conatives are set apart by its lack of result entailment and noncompletive interpretation (Krifka1999). Also in OSp, oblique realization of the object correlates to an interpretive shift from telic (5)b to atelic (5)a, reflecting the usual distinction that relies on realization of the object in this alternation (LR 2005:212). In fact, we find that all attested telic occurrences of ingestion verbs crucially lack the preposition. For instance, according to CES and SM data, conative variants combine with cessar de ?stop? but not with acavar de ?finish?, which is the prototypical pattern of atelic predicates. a. Asento se sobre el canto et el buitre cesso de comer en la molleja (15th c., CES) seated se over the pebble and the vulture ceased of eat in the gizzard ?He sat on the pebble and the vulture stopped eating at the gizzard? b. El solo comiesse el pan he alone eat the bread ?He only ate the bread? In COSL, alternation between direct object/oblique (PP) expressions in OSp also reflect alternation between telic/atelic uses. Like (5)a, (6)a is natural as the conative only entails that an unspecified (nonquantized) amount of people was killed (cf. causative/transitive entailing total affectedness, hence telicity (6)b). Similarly, in (3)a the endpoint PP headed by hasta imposes an event boundary that is otherwise missing (cf.(6)a). It follows that, even if the P used in OSp is different (linked in English to another (body-part-poss) alternation, Levin 1993), it still yields the event-type shift that sets CA apart (van Hout 1996 i.a.). In this sense, the selectional patterns attested in OSp (endpoint PP) are indicative of the expected contrast (cf. killed #(at) them as many as they could). In VERBS OF CONTACT, OSp alternation also yields a similar change in aspectual (Vendler 1967; Dowty 1979) structure. The generalization proposed for English holds here, as OSp conatives (i) do not necessarily involve a result state ((7) allows for the entailment that contact did not obtain as a result (He shot at them, but didn?t hit them)); and (ii) unlike Modern Spanish [MSp] prepositional variants (Acevedo 2011), they succeed in delivering the missed contact entailment (Dixon 1991) expected from hit-conatives (cf. MSp golpeó a la puerta #pero no la tocó ?He hit at the door but didn?t touch it?), otherwise (7) would be odd. In turn, productivity in (3)-(4) suggests that MOTION+ CONTACT denotation may not be necessary conditions for CA, or at least that the widelyclaimed restriction on conatives (LR 2015i.a.) may be language-specific (hold for e.g. English). (6) a. mataua en ellos quanto alcançaua. (14thc SM) kill.PST.IMP.3S in them how-much sufficed ?He killed at them as many as he could? Los sieruos que mataron a sus señores. (15thc SM) ?The servants that killed their lords? Y ordenó el Emperador que disparasen en los enemigos para que se apartasen. (16thc SM) and ordered el Emperor that shoot.SBJ.3P in the enemies for that se move-away ?And the Emperor ordered them to shoot #(at) the enemies so that they would move away? MAIN VARIATION: For COSL, we propose that in OSp verbs suggesting a causative/conative alternation differ in the realization of the external argument, instantiating a Cause(r)-Agent opposition amenable to a distribution linked to distinct v heads in the literature: while vDO requires an animate agent subject, the subject licensed by vCAUSE need not be (Folli & Harley 2005, 2007). We suggest that this difference in realization of the external argument can be seen in the OSp conative with causative verbs like matar: whereas the external argument in the causative only needs to be interpretable as cause(r) (La ponçoña lo mato ?The poison killed him?(15thc SM)), the (null) external argument of matar in the conative ((3)a, (6)a) requires an intentional agent, as in all attested cases. If correct, the analysis for conative (8)a vs. causative (8)b alternates could bear on composition with distinct v heads. This would capture the distribution drawn by restricted subject (agent) interpretation in CA, vs. causative-alternation instances of the same verbs. It follows that, apparently, both derivational paths are available for these verbs at least in OSp. New data (Múgica 2018) from (modern) Argentinian Spanish supports this possibility, as a number of verbs show all the hallmarks of a conative-structure-like derivation (result/telicity drop, missed contact entailment). Yet, while (8)a is fully and transparently productive in OSp, in MSp, (8)a is systematically used in incorporated forms (Mateamos por horas ?We drank at the mate for hours?). a. [vP [DP1 ,v? [vDO , PP [TARGET en DP2]]]] b. [vP [DP1 ,v? [vCAUSE, SC [DP2 , RES√]]]]