Alternative derivations of deadjectival verbs in Romance and measuring-out effects

MANGIALAVORI RASIA, MARÍA EUGENIA

Bayonne

Congreso; 25th Colloquium of Generative Grammar; 2015

IKER-CNRS

Alternative derivations of deadjectival verbs in Romance and measuring-out effects. The theoretical backdrop of this paper dwells on two different but not disconnected issues: (i) the relevance of the semantic content of property-denoting lexical roots (i.e. scale boundedness) triggering a root-event homomorphism shared by most languages; (ii) the alternatives following from the availability of different Path-encoding p0 heads for the derivation of deadjectival change-of-state [COS] verbs in Romance. Specifically, we will claim that semantic underpinnings dwelling on the open/closed nature of the property scale can truly determine the grammatical behavior (and structure) of verbs derived from property-denoting roots; but that, at the same time, this is a systematic though not generalizable phenomenon. Rather, we will show that this measuring-out effect depends on the type of little p head (p0) available for verbal derivation and the position in which the in the root is accommodated within the VP. Eventive classes and Event-root homomorphism. Based on the distribution of deadjectival verbs [DV] drawn by eventive tests such as ±compatibility with proportional/maximality modifiers and intensive modifiers, two (by this time long-claimed) classes of COS DVs can be distinguished. (1) embellecer ?embellish?, empeorar ?get worse?, empobrecer ?grow poorer?, endulzar ?sweeten? {#enteramente ?entirely?/ #parcialmente ?partially?/ #absolutamente ?absolutely?/ #completamente ?completely?/ #totalmente ?totally? / apenas ?barely?/ mucho ?a lot?/ demasiado ?too much?/ un poco ?a little?/ extremadamente ?extremely?} (2) emblanquecer ?turn white?, enverdecer ?turn green?, palidecer ?turn pale?, enloquecer ?go crazy?, enmudecer ?become mute?, ennegrecer ?turn black?, enrojecer ?turn red?, enronquecer ?grow hoarse?, ensordecer ?go deaf?, enturbiar ?become murky? {enteramente ?entirely?/ parcialmente ?partially?/ absolutamente ?absolutely?/ completamente ?completely?/ totalmente ?totally? / #apenas ?barely?/ #mucho ?a lot?/#demasiado ?too much?/ #un poco ?a little?/ #extremadamente ?extremely?} Scales and property-denoting roots. We ascribe the difference observed above to the root based on different observations. First, verbs in (1)-(2) share the same derivational constituents; moreover, the involvement of alternating affixes (other DVs use ?a/ø) is not relevant to the aspectual divergence. Second, both types of DVs can be accounted for on the basis of the same (l-syntactic) structure (e.g. conflation/incorporation forced by the phonologically empty v0, cf. Hale & Keyser 2002) and/or categorization processes. Third, the divergent patterns pointed out above do not follow from different categorial types of roots (they are all alternatively realized as Adjectives in Spanish). Finally, the mismatch does not correlate with diverse argument structure realization patterns: verbs from both (1) and (2) enter the causative alternation together, whereas other DVs, from both classes also, do not: some are inexorably transitive, others only feature an ergative layout. On the other hand, the divergent patterns in (1)-(2) can be readily explained as soon as the scale associated with the property-denoting root is considered, the simplest assumption about possible scale structures being that scales may or may not have maximal and minimal elements [±max]. Hence, whereas some properties seem to be associated to a bounded [+max] scale, others do not, therefore determining the eventive structure of the verb in accordance. As a consequence, we should expect DVs derived from [+max] roots to be telic and resultative, whereas DVs derived from [-max] roots shall render atelic and nonresultative COS. In in the latter case, the property associated with the root would not be interpreted as a result state, but just as the vector along which the (degree of) change can be measured. In fact, and as expected, the [+max] group succeeds in accommodating proportional and maximality modifiers (2), whereas the [-max] group handles the intensive modifiers rejected by [+max] roots (1). Therefore, as the division attested above is explained straightforwardly by [±max] degree on the scale, an event-root homomorphism (in the spirit of Harley 2005 i.a.) can be claimed rather straightforwardly. Further diagnostics consistently confirm this correlation. Do roots always measure out? Yet, this may not be a general or defective phenomenon, but a consequence of the type of path-encoding p0 used in the Spanish DVs comprised in ((1)-(2)). Data from Catalan and Italian lead us to observe that the measuring-out effect exerted by the root (i.e., measuring-out of the COS event via an event?root homomorphism) is actually sensitive to the position of the root within the DV. In particular, these languages offer two productive path-encoding components though which COS verbs can be derived: DVs can be either formed with an em? prefix (3) ?which is similar or equivalent to the Spanish affix in (1)-(2)?, or with an alternative path-encoding suffix (?eggiare/?ejar ) (4). What is appealing in this respect is that the resulting verbs, building on the same property-denoting root, show different aspectual properties. For instance, DVs derived from color-denoting roots should be resultative/telic, as generally agreed according to the bounded nature of the (scale imputed to the) root. Yet, pairings such as (5) show that a bounded [+max] root delivers a telic predication (i.e., it gets to measure out the COS event) only when it emerges after the path component (v0-p0-√ layout), but not when it precedes it. In fact, note that the ?eggiare/?ejar alternatives ?featuring the v0-√-p0 ordering? are not telic nor resultative even when they bear the same (color-denoting) root, typically described as [+max]. Hence, unlike en?DVs, which are defined as ?becoming x?, where x amounts to the property denoted by the root, ?eggiare/?ejar DVs are defined by denoting a never-ending change of state, essentially described as advancing towards (though never reaching) the property denoted by the root (cf. Oltra & Castroviejo 2013). Specific diagnostics support the proposed relation. Proportional and maximality modifiers are not possible with ?eggiare/?ejar DVs, but natural with en?DVs; in turn, intensive modifiers are out with the latter and possible with the former. Consistent facts are observed when different temporal /aspectual inflections are tested (e.g. past perfect tenses are only possible with en?DVs and imperfect ones are natural with ?eggiareDVs), or also with culminativity environments (6). (3) a. arrossire, ingiallire, sbiancare, inverdire b. enrogir, engroguir, emblanquir, enverdir (4) a. rosseggiare, gialleggiare, biancheggiare, verdeggiare b. vermellejar, groguejar, blanquejar, verdejar (5) a. {*abbastanza/*troppo/completamente/totalmente/parzialmente} inverdito quite too completely totally partially turn-green-PAST b. {abbastanza/troppo/*completamente/*totalmente/*parzialmente} verdeggiato quite too completely totally partially turn-greenish-PAST (6) Dopo aver finito di {arrossire/*rosseggiare} after have-INF ended of redden turn reddish More importantly, Italian provides a further insight on the (l)syntactic structure of these two alternative derivations, as long as auxiliary selection is admitted as relevant evidence. To be more specific, note that ?eggiare DVs consistently take avere ?have? as auxiliary, whereas en?DVs take essere ?be?. Among other things ?and if it is true that unaccusativity and ergativity do correlate with avere and essere selection, respectively? this finding is significant in the sense that the DP would not be in a measuring-out (internal) position in ?eggiare DVs either, thus contributing to the inherent atelicity/unboundedness of the event. Summing up, the two alternative realizations comprised by Romance DVs provide additional empirical evidence in favor of previous claims about measuring-out effects being sensitive to the position of the element exerting it (as proposed by Harley 2005i.a.). On the other hand, the alternatives comprised by Italian DVs also contribute non-trivial patterns connected with the relation between measuring-out and ergativity. Anyhow, Romance data consistently suggest that the scalar specification of the root may determine the eventive type of the VP, though this largely-studied phenomenon is actually subject to the formal properties of the heads available for conflation and the subsequent situation of the root within the (l)syntactic configuration of the DV. Last, but not least, the relation between the rightward position of the path-encoding element and the determination of telicity would involve a potential correspondence with patterns such as those observed by Koopman (2007) i.a.