Verbal telicity, lexical roots and VP structure. Notes on denominal and deadjectival verbs

MANGIALAVORI RASIA, MARÍA EUGENIA

Frankfurt

Congreso; Going Romance; 2016

Johann Wolfgang Goethe-Universität

Verbal telicity, lexical roots and VP structure. Notes on denominal and deadjectival verbs.Lexical roots [√] have been long argued to play a crucial role in verbal Aktionsart. Inspired by semantic work (e.g. Hay et.al. 1999), a considerable body of evidence is mounting that the aspectual (scalar) specification of the (thing/property-denoting) base can have a significant impact in the aspectual setup of denominal [DN] and deadjectival [DA] verbs, especially determining telicity. Previous work from the constructionist camp (Harley 2005) builds on the assumption that √s are inherently ±bounded to show that variable telicity of DNs/DAs can be reliably predicted from a systematic root-to-event relation. As the event is arguably modelled based on the scalar structure of √, the issue is analysed as Event-Root Homomorphism [ERH] (also Wechsler 2005).In the original proposal, Harley (2005) focuses on the observation that DNs produced by incorporation of bounded (count) noun roots are telic, whereas verbs produced by unbounded (mass noun) roots are atelic (1). She appeals to Hale & Keyser?s [HK] (2002) l-syntax to assert that, since √s originate in the object position of the (null) verb, they get to measure out the event in the same long-noted way incremental themes determine verb telicity in the (non-incorporated) transitive form (starting from Verkuyl 1972, i.a., Declerck 1979, Dowty 1979, Krifka 1998, Tenny 1987, Jackendoff 1991). This relation is proposed to extend to other DNs, including a type that figures prominently in HK?s original analysis; i.e., location/locatum verbs (2)? curiously enough, presented as a subtype of DAs (3)?, ERH holding for all of them. Given the evident structural difference with (1), ERH is more broadly defined as a direct (homomorphic) relation between V and the element sitting in sister-to-v position (?the SC itself? in (3) 2005:54). As the verbs paired in (1)-(3) are (admittedly) configurationally alike, and eventual (a)telicity hinges on the content of the root, then telicity is claimed as an automatic non-configurational result. Now, given that comparable patterns appear in other languages, Romance included, ERH can be crosslinguistically deployed in the prediction of variable telicity DN and DAs. Yet, a more extensive analysis of data, especially of Italian DNs and DAs suggests that ERH does not generalize so easily. a. {The mare foaled/The dog whelped/The cow calved} {#for two hours/in two hours}b. {The baby drooled/The athlete sweated/The wound bled} {for two hours/#in two hours} a. Bill {saddled the horse/boxed the computer} {#for two hours/in two hoursb. Bill {buttered the bread/greased the chain} {for two hours/#in two hours} a. Bill {cleared/flattened/dried} the screen {#for two hours/in two hours}b. Bill {lengthened/widened/darkened} the screen {for two hours/#in two hours} Problem #1. Crucially, the examples in ((2)b)?i.e., the case where unbounded nominal roots yield atelic verbs?only feature locatum verbs. Location DNs from the original set (HK 2002:18) on unbounded roots are, unexpectedly for ERH, telic (4)?and so are their Romance equivalents; cf. It. affondare, aterrare, ammarare. Analysis. Even if there is a growing tendency to analyze locatum/location as a difference defined conceptually/encyclopedically (Mateu 2001, Harley 2005 i.a.), an alternative building on P flavors would readily account for (4). Abiding by HK?s original account, analyzing locatum/location verbs as the result of a grammatical alternative (i.e., recruiting a particular type of P), it could thus be argued that ERH is sensitive to the ?fundamental semantic properties of the Ps involved? (2002:19). Namely, if the specific properties of P (e.g. a Terminal-Coincidence type of P) determines that √ is interpreted as Terminal Ground [TG], telicity is expected to arise automatically, regardless of the √ type; instead, if √ is mapped as an entity coinciding with the external argument in a possessive relation (locatum), different outcomes are possible (e.g. locatum use of floor has a stative reading: The room is floored in oak) and potentially predicted by ERH according to the type of √ involved (with object measuring-out remaining active as secondary variable). Given the invariable telicity of location DNs, regardless of the ±unboundedness of √ (e.g. mass root), variable telicity would not be directly determined by √ as a non-configurational result, but rather sensitive to the l-syntactic structure hosting √. Bill {landed the plane/grounded the ship/floored the pedal/skied the rocket}{#for/in two hours} Problem #2. As for DAs, the picture is more complex, but the divergences at stake are quite as relevant. As we showed in previous work, the essential patterns revealed by English and Romance DAs (3) give empirical support to the hypothesis that verbal telicity follows directly from thescalar structure of the property-denoting base. Specifically, distribution of partial/total adjuncts (Hay et.al. 1999 i.a.) concurs with further tests indicating that bounded property scales reflects not only in ±telicity, but in ±resultativity and ±homogeneity as well. The boundedness of the DP (object measure-out), though relevant, has only a secondary effect, as the scale contributed by √ remains the crucial determiner. On these grounds, ERH bears out. But then, a further set of DAs available in Romance (productive in Italian, Catalan [see Oltra&Castroviejo 2013], and European Portuguese) is shown to consistently produce atelic verbs, even if building on bounded (property-denoting) roots producing telic verbs in (3). Notably, √s associated with bounded scales like pian- [flat], rotond- [round], bianc- [white] produce telic DAs of the type in ((5)a) and invariably atelic verbs in the type ((5)b). AnalysisMajor patterns show that this second type of verbs (-eggiare/ejar) are not only produced by different morphology, but actually involve different eventive and argument structure configurations (ne-cliticization, auxiliary distribution, causative/inchoative alternation, etc. consistently show -eggiare verbs to be stative and unergative, in contrast to the unaccusative COS verb in (5)a)). If correct, the role of √ in verbal telicity (hence, of ERH in DAs) is then relative to the semantic and syntactic structure of the VP hosting it. Finally, if we extend the analysis to DNs formed by the same morphology, results show the same argument structure (unergative) and eventive type as DAs for which ERH does not hold, but also a consistent atelicity even if featuring bounded thing-denoting roots (6). Crucially, they test out as stative even if often paraphrased by activity verbs, and do not involve either change of state nor location, nor √ is interpreted as incremental theme (unlike DNs in (1)) (7). Thus, delivery of telic/atelic verbs would be strictly configurational (i.e., the result of different syntactic and semantic structures) here. a.{#scaldare/appianare/#ammollire/arrotondare/sbiancare} {in due ore}heat [up]/flatten/ soften round [up] whiten in two hoursb. {#caldeggiare/#pianareggiare/#molleggiare/#rotondeggiare/#biancheggiare} {in due ore}be white(ish)/ be flat(ish) be tender(ish) be round(ish) be white(ish) in two hours {americaneggiare/ berlusconeggiare/ bambineggiare} {#in due ore/per ore}look.american(ish) resemble Berlusconi resemble a child in two hours for hours a. {boccheggiare/ bordeggiare/ barcheggiare/ fondeggiare/ mareggiare} {per ore}mouthv / (remain by)border/boat [around]/stay over seabottom/move like the sea for hoursb. {imboccare/ abbordare/ imbarcare/ affondare / ammarire} {#per ore}put in mouth board embark sink land at sea for hours Extensions. If we focus on argument structure, the fact that v and √ are mediated by P in (2)-(4) can be pushed further to propose that locatum/location DNs and DAs would be a case, not of measuring-out but of delimitness (Tenny 1994:11). Whereas ERH would be a form of event-object homomorphism in (1); (2)-(4) could be recasted as a further instance of Event-Path Homomorphism, insofar as telicity corresponds a trajectory across bounded (spatial/property) ranges (given a holistic effect), in contrast to the stative DA (5)b-(7)b), and the role of the affected object (measuring-out) is only secondary. If on track, ERH could be seen as an epiphenomenon of two well-known, but crucially different, argument structure mapping generalizations.  Conclusions The premise that telicity is directly determined by the semantic (scalar) specification of the lexical root under the ERH model is broadly consonant with conclusions reached independently in previous works and explains major empirical patterns in Romance verbs: (i) DNs/DAs featuring the same l-syntactic structure render telic verbs iff √ is associated to a bounded scale; (ii) √ is the locus of variation/determination (either encyclopedically or lexically) of non-trivial semantic properties (iii) telicity is non-configurational. Yet, the generalization previously adduced faces two empirical challenges: (#1) DN (location) verbs yield invariable telicity (with ±bounded √s); (#2) Romance DN/DA yield invariable atelicity (with both ±bounded scales). These two gaps show that ERH is visibly constrained. In DNs (Gap#1) ERH may be non-configurational but depends on the semantics of a closed-class element (P). Gap (#2) shows that for DAs variable telicity is directly determined by √ only in certain configurations (cf. (5)a-b); in this sense Romance DAs show that variable telicity in DA/DN can also be a configurational result.Sel. References: Hale, K. & S. Keyser. 2002. Prolegomenon to a theory of argument structure. MIT ? Harley, H. 2005. How do verbs get their names? Erteschik, N. 2005. The syntax of aspect. OUP ?Hay, J; C. Kennedy & B.Levin.1999. Scale structure underlies telicity in ?degree achievements?. SLT 9.127?144? Tenny, C. 1994 Aspectual Roles and the Syntax-Semantics Interface. Kluwer, Dordrecht.