An overlooked causative variant? Revisiting the causative alternation in Romance

MANGIALAVORI RASIA, MARÍA EUGENIAIA

Pavia

Congreso; 43 IGG; 2017

USS - Istituto Universitario di Studi Superiori di Pavia

Transitivity alternations are a fundamental issue in the study of the lexicon/syntax interface. If we standardly assume that the dyadic frame in (1) corresponds to a complex event structure, in contrast to the simpler event associated with the monoargumental variant (see (2)), an important fact about the causative-inchoative alternation is that the structural contrast (and variable complexity) in argument realization coincides with a contrast (and relative complexity) bearing on event structure.(1)a. John closed the door [CAUSEX[BECOME[Y]]] b. The door closed [BECOME[Y]](2) VTRANSITIVE = CAUSE to VINTRANSITIVE (Levin 1993:27)Given that lexical overgeneration is not a desirable upshot, a relevant question becomes whether these variants derivationally related; and, if so, which structure is the ?original? and which one is derived. While a wide spread of accounts analyze the relative complexity of the transitive variant as a composition combining the causative structure with a basic (inchoative) structure (e.g., Dowty 1979, Parsons 1990, Hale & Keyser 1993, Rappaport & Levin 2011); others propose instead that the transitive variant is basic and the inchoative is derived (Levin & Rappaport 1995, Reinhart 2002, Chierchia 2004 i.a.). More recently, non-derivational approaches (Rosen 1996, Doron 2003, Mateu 2000, Harley 2008) argue that different argument structure configurations can be produced from the same root. In any case, a major theoretical discussion holds between works placing the burden of the explanation in the lexicon (Levin & Rappaport 1995, Jackendoff 1990, Reinhart 2002) and works proposing that it should rather be placed in the syntax (Borer 2005, Travis 2000, Kratzer 1996, Marantz 1997, Ritter & Rosen 1998, Harley 1995, among others), which is taken to include the composition of ?lexical? argument structure configurations (Hale & Keyser 1993, 2002, 2005).1.Problem. A fact which is not commonly discussed from the perspective of transitivity alternations is that the variability in the syntactic frame arguably extends to a third alternative in Romance, also related through the notion of "cause". This less-visited variant features a single argument, which is defaultly interpreted, not as the undergoer (i.e. as in (b)), but rather as the entity with the relevant properties to cause the change (i.e., the initiator) ((3)c)?i.e., the DP2 in the transitive form. Interestingly, even if this monoargumental (initiator-only) alternative is acceptable in English in some contexts, it seems to be quite free in Romance (encyclopedic restrictions aside) (cf. (3)-(4)). (3)a. El cloro blanquea los huesos.?Bleach whitens the bones?b. Los huesos (se) blanquean. ?[the] Bones whiten?c. El cloro blanquea. ??Bleach whitens?(4)a. La comida chatarra engorda a los niños.?Junk food fattens the kids?b. Los niños engordan. ?Kids fatten [up]? c. La comida chatarra engorda.*?Junk food fattens?The construction in question, called here stative-causative [SCC] to reflect its semantic and eventive status (see also Rothmayr 2009:47), creates a problem for the structural characterization of change-of-state and deadjectival verbs in at least three respects. First, SCCs raise the question as to whether the internal argument?which in fact appears as the only apparent constant in change-of-state verbs when the analysis is limited to (1) or to (a/b) cases in (3)?is indeed a default constituent in the argument structure configuration of these verbs, as commonly assumed in both lexicalist and constructional approaches (cf. Ramchand 2008 for summary). Second, by suggesting that the causative component may be realized independently, SCCs challenge a basic rule of event composition that figures in lexicalist and constructionalist approaches, whereby the event structure of change-of-state verbs combines two basic components in a hierarchical relation, cause and process, and that the former, when present, causally implicates the latter (Hale & Keyser 1993:69, Rappaport Hovav & Levin 1998:108). Third, the consistent unergative behavior of SCCs suggests that verbs which are commonly classified as bona fide unaccusatives (cf. Ramchand 2008:35, Levin & Rappaport 1995:80) or as basic transitives (e.g. Levin 1993, Levin & Rappaport 1995:25) may yield unergative frames or at least have unergative uses in Romance. 2.Proposal We will treat this particular kind of flexibility on structural alternations by proposing that it depends on the type of verbal head (vo) combined in the formation of the verb, rather than of the properties of a particular lexical entry, with three potential configurations resulting from a common (perhaps category-neutral) root. We argue that the possibility to derive ((3)/(4)c) can be explained by a (null) causative vo which is freely available in Romance for derivation, producing a simpler event and argument structure than the one seen in ((3)/(4)a), but at the same time, one which is crucially (semantically and syntactically) different from the simple structure seen in ((3)/(4)b). Importantly, the transparent correlation between argument and event structure is preserved: based on specific tests and independent data collected via corpus search and experimental surveys, we will show that in SCCs the absence of the internal argument correlates with the absence of the process (sub)event, producing a stative predicate with unergative behavior. Notably, SCCs do not tolerate progressive/perfect tenses or endpoint adverbials; further, if in-X-time adverbials, degree modifiers, and locative PPs occur at all with SCCs, they deliver the interpretation expected for statives (e.g., measure the extent of the initiational capacity (property)). A different but nonetheless interesting problem is that the stativity of SCCs contrasts with stative predicates analyzed as a result of a causative vo (vCAUS/INIT) in previous accounts (e.g. English Katherine fears nightmares, Ramchand 2008) in two crucial ways. First, in SCCs the subject is interpreted by default as a causer or trigger rather than as holder of a result state. Second, the potential unergativity of the construction?which follows straightforwardly from the structural properties of the causative vo, which characteristically licenses external arguments?is not predicted by the standard analysis, and contrasts with the argument structure realization patterns displayed by English (e.g. X fears/annoys Y). To solve this, we propose that a non-derivational (constructional) approach, using distinct and meaningful v heads to constrain interpretation and argument selection, provides a principled explanation for the problem. Although a specific type of (l-syntactic) account will be implemented in the argumentation (v flavors, Folli & Harley 2007), direct composition with voINIT (Ramchand 2008, McIntyre 2004), we will work under the more general assumption that the semantic properties of the different verbal heads combined are responsible for event interpretation and argument structure realization, in line with earlier works (cf. Hale & Keyser 1993, Borer 2005:30). 3. Advantages. ?The absence of a process component in initiator-only frames supports a transparent correlation between the simpler event structure of SCCs (state) and the simpler (unergative) syntactic frame, thus reinforcing the idea of an internal-argument-licensing process component (e.g. vBECOME) not included in the representation by default. This in contrast to an explanation building on implicit arguments and/or unrealized projections which is not favored by empirical data?unlike null object constructions, SCCs do not allow bare quantifiers (5), reflexive morphology (6), secondary predication (7), ne-cliticization (Italian)?and is consistent with similar observations concerning atransitivity (cf. McIntyre 2004(59)). ? The stativity of SCCs would be consonant with the stative nature of the causative vo suggested in recent constructional approaches (Ramchand 2008 i.a.), while its atelicity agrees with the general principle (going back to Dowty 1979) that relates unergativity/atelicity on the one hand, and unaccusativity/telicity on the other hand. ? The constructional status of the process event, together with the corresponding verbal projection, fits well with the well-known empirical observation that unaccusative frames are morphologically complex in Romance?a powerful argument commonly adduced by proponents of a causative-to-inchoative derivation (see Ramchand 2008 and Rappaport & Levin 2011 for summary). ? The analysis of SCCs as the result of free compounding with a causative vo (voCAUS, √) is consistent with the significantly fewer cases of change-of-state/deadjectival verbs resisting a causative form. ? The fact that the addition of the causative component is apparently only constrained by external semantic conditions agrees with arguments from more recent constructional (e.g. Harley & Noyer 2000, Hale & Keyser 2005) and lexicalist (e.g. Rappaport & Levin 2011) views on this topic. (5)Esto {enfurece/empobrece/indigna}*(muchos). ?cf. Esto ha dejado muchos indignados/pobres.?This infuriates/impoverishes/outrages many? ?This has left many infuriated/poor?(6)La injusticia {entristece/enoja} (*con sí mismo). ?cf. (se/lo) enoja/entristece (con símismo).?Injustice saddens/maddens *with oneself? ?(It) saddens/maddens one with oneself? (7)El horno calienta (*congelado/*abundante/*crocante). ?cf. El chef cocina congelado/rico.?The oven heats packed/compact/crispy? ?The chef cooks frozen/tasty [meals]? Selected ReferencesLevin, B. 1993. English Verb Classes and Alternations. University of Chicago Press.Hale, K. & S. Keyser. 1993. ?On the argument structure and the lexical expression of syntactic relations?. In The View from Building 20. MIT Press. Levin. B. & M. Rappaport. 1995. Unaccusativity. MIT Press.Ramchand, G. 2008. Verb Meaning and the Lexicon. Cambridge University Press.Rappaport, M. & B. Levin. 2011 ?Lexicon uniformity and the causative alternation?. In: Everart, M., M. Marelj & T. Siloni, (eds), The Theta system. Oxford University Press.McIntyre, A. 2004. ?Event paths, conflation, argument structure and VP shells?. Linguistics 42(3) 523-571.